Cognitive Neuropsychology

ISSN: 0264-3294 (Print) 1464-0627 (Online) Journal homepage: https://www.tandfonline.com/loi/pcgn20

Atypical holistic processing of facial identity and
expression in a case of acquired prosopagnosia
Cecilia Monti, Matteo Sozzi, Francesco Bossi, Massimo Corbo & Davide
Rivolta
To cite this article: Cecilia Monti, Matteo Sozzi, Francesco Bossi, Massimo Corbo & Davide
Rivolta (2020): Atypical holistic processing of facial identity and expression in a case of acquired
prosopagnosia, Cognitive Neuropsychology, DOI: 10.1080/02643294.2020.1718071
To link to this article: https://doi.org/10.1080/02643294.2020.1718071

Published online: 27 Jan 2020.

Submit your article to this journal

Article views: 23

View related articles

View Crossmark data

Full Terms & Conditions of access and use can be found at
https://www.tandfonline.com/action/journalInformation?journalCode=pcgn20

COGNITIVE NEUROPSYCHOLOGY
https://doi.org/10.1080/02643294.2020.1718071

Atypical holistic processing of facial identity and expression in a case of acquired
prosopagnosia
Cecilia Montia, Matteo Sozzia,b, Francesco Bossic, Massimo Corboa and Davide Rivoltad
a

Department of Neurorehabilitation Sciences, Casa di Cura del Policlinico, Milano, Italy; bDepartment of Neuroscience, Neurology Unit,
“A.Manzoni” Hospital, Lecco, Italy; cIMT School for Advanced Studies, Lucca, Italy; dDepartment of Education, Psychology and Communication,
University of Bari Aldo Moro, Bari, Italy
ABSTRACT

ARTICLE HISTORY

Typical face perception is mediated by holistic processing (i.e., the simultaneous integration of face
parts into a whole representation). People with Acquired Prosopagnosia (AP), who have lost the
ability to recognise faces after a brain lesion, should thus show atypical holistic coding. Our aim
is to use the composite-face effect (CFE) as a measure of holistic processing in ST, a 48-year-old
woman with AP but normal recognition of facial expressions of emotions, and matched healthy
control participants. Two experiments examining the CFE for identity (Experiment 1) and for
expression of emotions (Experiment 2) were conducted. Contrary to controls, in both
experiments, ST showed an atypical (i.e., reversed) CFE, thus suggesting altered holistic
mechanisms affecting both components of perceptual judgement. Results also suggest that
normal facial expression recognition is achievable even with holistic processing difficulties,
possibly through compensatory, part-based, mechanisms.

Received 8 August 2018
Revised 2 January 2020
Accepted 7 January 2020

1. Introduction
Cognitive and neural evidence strongly indicates that
faces represent a special category for the visual
system. Human face recognition mostly relies on the
synchronous activity of dedicated neurons in the
(mostly right) occipito-temporal cortex (Haxby,
Hoffman, & Gobbini, 2000; Kanwisher et al., 1997;
Rivolta, Castellanos et al., 2014a; Rivolta, Woolgar
et al., 2014b). Lesions of occipito-temporal
regions cause acquired prosopagnosia (AP), a neuropsychological condition characterized by the impairment in recognizing previously known faces, and the
inability to form memories of new faces despite the
absence of low-level visual defect or intellectual disorders (Bodamer, 1947; Damasio, Damasio, & Van
Hoesen, 1982; Grüsser & Landis, 1991; Mayer &
Rossion, 2013; Rondot & Tzavaras, 1969). The recognition of familiar people in AP mainly relies on extrafacial cues, such as hair, accessories, gait, clothes and
voices (see Benton, 1990 for a review).
Studies from cognitive psychology demonstrate that
typical face recognition relies on dedicated perceptual
mechanisms known as holistic, which refer to the simultaneous integration of face parts into a whole

KEYWORDS

Face; identity; expression;
emotions; holistic;
prosopagnosia

configuration (Mckone & Yovel, 2009). The idea that
holistic coding may allow for efficient face representation (Maurer, Le Grand, & Mondloch, 2002; Piepers
& Robbins, 2013) was supported by studies demonstrating correlations between measures of holistic processing (see below) and face memory skills in the typical
population (see DeGutis, Wilmer, Mercado, & Cohan,
2013 for correlations between holistic measures and
face memory; and Richler et al., 2011 for the role of holistic processing in predicting face-recognition abilities),
and by the aberrations of holistic mechanisms in
patients with AP (e.g., Busigny, Joubert, Felician, Ceccaldi, & Rossion, 2010, 2014; Ramon, Busigny, &
Rossion, 2010; Van Belle, de Graef, Verfaillie, Busigny,
& Rossion, 2010). One of the most reliable measures
of holistic processing is the composite-face effect (CFE),
a visual illusion where the top half of one (upright)
face appears to fuse perceptually with the bottom
half of another, in the condition where the two halves
are aligned (as opposed to when they are misaligned)
(Hole, 1994; Young, Hellawell, & Hay, 1987). Albeit atypical CFE in AP has been documented in most cases
(Busigny et al., 2010, 2014; Ramon et al., 2010), the existence of APs showing the typical CFE (Finzi, Susilo,

CONTACT Davide Rivolta
davide.rivolta@uniba.it
Department of Education, Psychology and Communication, University of Bari Aldo Moro, Bari, Italy
Supplemental data for this article can be accessed https://doi.org/10.1080/02643294.2020.1718071
© 2020 Informa UK Limited, trading as Taylor & Francis Group

2

C. MONTI ET AL.

Barton, & Duchaine, 2016; Rezlescu, Pitcher, & Duchaine, 2012) indicates that face recognition deficits can
co-occur with intact holistic processing (which might
be due to the extent and site of the brain lesion).
Note, however, that different types of holistic mechanisms exist (Mckone & Yovel, 2009; Negrini et al., 2017),
which are sensitive to different manipulations (e.g.,
the part-whole effect, Tanaka & Farah, 1993; the faceinversion effect, Yin, 1969), and that could indicate atypical holistic mechanisms where the CFE fails to find
group differences.
Albeit face identification tasks and the CFE often
show anomalies in APs, more variability is seen in the
recognition of expressions of emotions. According to
standard models (Bruce & Young, 1986) these two abilities can dissociate; there are, in fact, APs with preserved
emotion recognition (Riddoch, Johnston, Bracewell,
Boutsen, & Humphreys, 2008; Takahashi et al., 1995;
McNeil and Warrington, 1993; Tranel, Damasio, &
Damasio, 1988) (but see Fox et al., 2011). The patient
FB (Riddoch et al., 2008) presented a severe acquired
prosopagnosia yet scored within the typical range on
an expression recognition test (i.e., The Emotional
Hexagon Test, Ekman & Friesen, 1976, which
implemented faces from the Ekman and Friesen set
that have been manipulated to produce images of
graded difficulty; FB score: 99/120; cut-off: 94). The
Authors suggested that FB relied on part-based information to a greater extent than controls and this
allowed the patient to correctly solve the task. This is
also seen in the congenital (or developmental) form
of prosopagnosia (CP), which refers to the lifelong
impairments in recognizing people by their faces
(Duchaine, Parker, & Nakayama, 2003; Rivolta, Schmalzl,
Coltheart, & Palermo, 2010; 2017). Despite face recognition difficulties CPs often (Humphreys, Avidan, &
Behrmann, 2007; Lee, Duchaine, Wilson, & Nakayama,
2010; Palermo et al., 2011; Rivolta et al., 2010), albeit
not always (Biotti & Cook, 2016), show typical
emotion recognition. Thus, AP and CP cases suggest
that holistic coding is necessary for identification but
not for expression, and that emotion recognition
might rely on featural, rather than holistic, coding
(Tanaka, Kaiser, Butler, & Le Grand, 2012). From the
theoretical point of view, a direct way to test this
hypothesis is to check whether APs show typical or atypical CFE for emotions; atypical CFE for emotions would
suggest that (atypical) holistic mechanisms for facial
expressions of emotions are not sufficient to cause

facial emotion recognition deficits (Palermo et al.,
2011). Unfortunately, this evidence is lacking in AP.
Thus, we here describe the case of ST, a 48-year-old
Italian woman who, following right occipital brain
damage, shows serious everyday-life difficulties in
identifying familiar faces, albeit with typical ability to
recognize facial expressions of emotions. The first
aim was to investigate whether the impairment in
face identification was associated with deficits in holistic processing (i.e., CFE for identity) (Experiment 1).
In addition, to test whether facial expressions of
emotions are processed holistically or via a partbased processing, we adopted the CFE for emotions
(Experiment 2) (Calder & Jansen, 2005; Calder, Keane,
Young, & Dean, 2000).
According to her neuropsychological profile (see
below), we hypothesized that ST was sensitive to holistic interference (i.e., CFE) when she had to judge
emotional expression, but not when the task required
judging facial identity. This pattern would suggest a dissociation between the two cognitive processes, which
arises earlier than holistic perceptual processing.
However, considering the evidence of an interaction
of emotion and identity mechanisms at low-level
stages of processing in healthy subjects (Fisher,
Towler, & Eimer, 2016), an alternative hypothesis
would be that ST had impaired holistic processing for
both emotion and identity judgements. If this was the
case, her spared ability to categorize emotions from
facial information despite severe difficulties in identity
recognition could be attributed to the use of compensatory strategies (feature-by-feature processing), or, more
broadly, it might suggest that holistic perception is not
essential for expressions recognition, which primarily
relies on part-based processing (Tanaka et al., 2012).

2. Methods and results
2.1. ST: case description
ST is a right-handed woman with 17 years of education (Master’s degree in economy). She was 48
years old at the time of the assessment. In September
2014, she sustained a right occipital haemorrhage
associated to thrombosis of the transverse and
sigmoid sinuses. TC scan obtained in October 2014
revealed the presence of a haemorrhagic infarct in
subacute phase in the occipito-temporo-parietal
region of the right hemisphere. Following, a structural

COGNITIVE NEUROPSYCHOLOGY

MRI showed the result of the intraparenchimal hematoma in the right occipito-parietal regions (Figure 1).
Consecutively, ST suffered from left homonymous
hemianopia and prosopagnosia. During her hospital
stay, ST complained of severe difficulties in recognition of familiar faces, such as relatives (e.g., her
husband, sons) and friends. ST did not report any
difficulty with face expression recognition. She lamented severe problems in recognizing familiar people,
which she described as even more evident in condition of no context cues and expectations (e.g.,
work colleagues met outside the work place). She
reported to rely on features (eye colour, earrings,
beard, etc.) to overcome her face identification
difficulties. ST also complained to sometimes see
faces as distorted, a condition known as prosopometamorphopsia (PM) (i.e., a rare face-selective disorder in
which faces appear distorted as they were drooping,
floating or bulging, Hecaen & Angelergues, 1962).
Specifically, she reported that known faces appeared
uglier and darker than she remembered. In the
weeks immediately following the acute event, such
misperceptions were persistent but not stable in
terms of intensity, since they were reported as
variant according to environmental conditions (i.e.,
brightness). In a second assessment 5 months after

Figure 1. Structural MRI showing ST’ lesion affecting the right
occipito-parietal regions (axial plane).

3

brain damage, ST noticed relevant improvements
reporting that faces appear less swollen and ugly
than earlier. At the 1-year follow up, the prosopometamorphopsia has been completely recovered.
2.2. The diagnosis of prosopagnosia
Immediately after the acute event ST complained to
experience severe everyday life difficulties in the recognition of previously known (e.g., familiars, relatives and
friends) and new (e.g., professionals met at the hospital)
faces. Nevertheless, she was able to recognize familiar
people by the voice, clothes, and non-identity facial features (i.e., beard and glasses). Both ST and her relatives
reported she never had difficulties in recognizing faces
before the acute event. ST reported such condition as
particularly disabling due to mistakes in recognition
of people who were even highly familiar to her, such
as her husband and sons. She was completely aware
and worried for such difficulties, which dramatically
affected her everyday life.
With the aim to formally assess ST’s face recognition
abilities, and to “diagnose” prosopagnosia, we administered three tasks: (1) The recognition of familiar faces
was examined by means of an experimental task
requiring identification of ST’s relatives, friends and colleagues; (2) the ability to learn new faces has been
assessed with the Cambridge Face Memory Test
(CFMT; Duchaine & Nakayama, 2006), which is one
the most adopted and valid test to diagnose prosopagnosia (Bowles et al., 2009); and (3) face perception has
been assessed with the computerized version of the
Glasgow Face Matching Test (GFMT, Burton, White, &
McNeill, 2010), a perceptual task that requires matching of same-identity, but perceptually dissimilar,
faces. Concurrently, an extensive neuropsychological
assessment was conducted in order to exclude the
presence of perceptual or high-order cognitive
deficits. ST’s performance was compared to that of
typical subjects by means of a t-test for small samples
(SINGLIMS t-test; Crawford, Garthwaite, & Porter, 2010).
2.2.1. Familiar face recognition
From the historical perspective, prosopagnosia has
been defined as the inability to recognize people
that were familiar before the traumatic event
(Hecaen & Angelergues, 1962). As such, we examined
ST’s ability to recognize 38 personally familiar faces,
which included her son, daughter, husband and

4

C. MONTI ET AL.

relatives (three pictures of the patient’s own face have
also been included). All extra-face details (e.g., hair,
clothing, jewellery) were cropped using the Open
Source software GIMP. Each stimulus was presented
on a white background until response. ST was asked
to identify each individual face by trying to name it.
In case of difficulty, we asked ST to describe whether
faces seem familiar or not. All ST’s verbalizations
were registered. Overall, ST correctly recognized 50%
of the shown faces. The patient showed a lot of uncertainties and a very long time to answer (over 3 sec).
She was not able to recognize her own face, which
was wrongly attributed to her father. Frequently, ST
reported a failure to feel a sense of familiarity;
however, when recognition was possible, naming
occurred almost immediately with no hesitations.
When asked, ST reported to rely on specific outstanding details, which were described as particularly
salient (i.e., shape of eyes, their colour etc.). Albeit
not specifically asked in the task, the patient also
showed difficulties in gender and age attribution:
she reported to rely on cues such as the presence of
wrinkles to describe the face as depicting old or
young person. As a control, ST’s husband has also
completed the task and was able to correctly identify
all the faces. Taken together, this finding strongly
confirmed the presence of prosopagnosia.
2.2.2. Unfamiliar face learning
The CFMT examines immediate memory for unfamiliar
faces, requiring learning of target faces from different
views. After a practice stage, subjects have to complete three phases: (i) introduction: recognition of six
previously learned faces among distractors (test
items with identical images); (ii) novel images: recognition of the target faces in which the test items are
novel images (i.e., the lighting, pose or both vary);
(iii) novel images with noise: recognition of the
target faces presented with different levels of Gaussian noise. ST’s accuracy (60%) was significantly
lower (t = −2.02, p = .02, one-tailed) than control data
(N = 20; mean: 82.7%; SD: 7.6; mean age: 45.1) published by Duchaine, Yovel, and Nakayama (2007)
(Table 1). This result suggests impairment in learning
and recognizing new faces.
2.2.3. Face matching
Face identification relies upon the fundamental ability
to perceive a face as the same even when it is seen

with different lighting conditions, distance and head
orientation (i.e., within-identity changes). Such ability
refers to the structural encoding process, which
allows for the construction of a view-dependent representation of the face (Bruce & Young, 1986). Here
we used the Glasgow Face Matching Test (GFMT,
Burton et al., 2010) in order to observe whether ST’s
difficulties in face recognition encompass the lowlevel phase of structural encoding. The GFMT (short
version) measures the ability in matching two simultaneously presented faces that have the same frontal
viewpoint but are subject to within-identity change
(White, Rivolta, Burton, Al-Janabi, & Palermo, 2017).
This introduces within-identity variations in the
aspect and metric distances between images of the
same individual. Each of the 40 face pairs (20 same
pairs; 20 different pairs) was shown until response.
On each trial, participants are required to indicate
whether the two images were of the same or of
different person by pressing the correspondent
key on a keyboard. The experiment was run with
E-Prime 2.0.
As compared with normative data (Mean = 81.2%;
SD = 9.7; N = 194; mean age: 26, age range: 18–46)
provided in Burton et al., 2010, ST’s accuracy performance (60%) was impaired (t = −2.250, p = .01, onetailed) (Table 1). RTs were not analysed since ST
showed excessively long time for response (mean
correct RTs: 9.3 sec). Overall, performance on the
GFMT suggests early, perceptual, problems in face recognition, possibly characterizing her prosopagnosia as
apperceptive in nature (De Renzi, Faglioni, Grossi, &
Nichelli, 1991).
Table 1. ST’s performance on standardized measures of face
processing and facial expression recognition as compared to
controls.
Face processing
CFMT
GFMT
Facial expression recognition
Ekman 60 Faces Test
Suprise
Happiness
Fear
Disgust
Anger
Sadness
Total score

Score

p value

43/72
24/40

.02*
.01*

Score
10/10
10/10
4/10
5/10
6/10
8/10
43/60

Cut-off
6
9
2
4
5
4

Note: CFMT = Cambridge Face Memory Test; GFMT = Glasgow Face Matching
Test.
* SINGLIMS one-tailed significant p values.

COGNITIVE NEUROPSYCHOLOGY

2.2.4. Facial expression recognition
ST’s ability to recognize emotions from facial
expressions was tested with the Ekman & Friesen 60
Faces Test (Young, Perrett, Calder, Sprengelmeyer, &
Ekman, 2002), which examines the recognition of six
basic facial expression (i.e., anger, disgust, fear, happiness, sadness, surprise). The test is composed of 60
trials preceded by a practice of 6 trials. Stimuli
belong to the Ekman & Friesen set of Pictures of
Facial Affect (1976) and appear on the screen for 5
sec each. Participants had to choose between six
emotion labels, which remain on the screen for the
same trial duration. Responses were verbal. ST’s
global score (raw score: 43/60; corrected score:
41.05/60, adjusted for age, gender and education) is
within normal limit (equivalent score = 1) as compared
to a sample of Italian subjects (N = 132, mean age
51.11, SD 16.5; Dodich et al., 2014) (Table 1).
2.2.5. General neuropsychological evaluation
A first neuropsychological assessment was conducted
three weeks after injury. ST’ spatial, temporal and

5

personal orientation were adequate. She performed
within the normal range in several tasks assessing
spatial cognition, praxis abilities, abstract reasoning,
executive functions, verbal long-term and short-term
memory and attention. Moreover, she performed
well in tests evaluating colour discrimination and recognition. See Table 2 for a summary of results.
Further, a detailed evaluation of low-level visual
perception and objects recognition abilities was conducted in order to investigate the presence of
objects agnosia, which is frequently described in
patients with AP (Clarke, Lindemann, Maeder,
Borruat, & Assal, 1997). First, the Leuven Perceptual
Organization Screening Test (L-POST) (Torfs, Vancleef,
Lafosse, Wagemans, & de-Wit, 2014) was administered
to screen for possible deficits in mid-level vision. The
L-POST is composed of 15 subtests that cover a
range of mid-level processes such as figure-ground
segmentation, local and global processing, and
shape perception (see Torfs et al., 2014, for a detailed
description of each subtest). ST’s overall score was
93.34% (70 out of 75 correct); the performance was

Table 2. Results of the general neuropsychological assessment.
Cognitive domain
Visuo-spatial abilities

Ideomotor apraxia
Buccofacial apraxia
Nonverbal executive functions
Numerical abilities

Test
Line bisection (horizontal) (Fortis et al., 2010)
Cancellation task—Bells Test (Gauthier, Dehaut, & Joanette, 1989)
Sentence reading (Zoccolotti et al., 1989)
Vertical reading
Line bisection (vertical)
Imitation of gestures (De Renzi, Motti, & Nichelli, 1980)
Imitation of orofacial movements (Spinnler & Tognoni, 1987)
Clock Drawing Test (Mondini, Mapelli, Vestri, & Bisiacchi, 2003)
E.N.P.A.
Numbers repetition
Dictation
Words-numbers
Calculations—sum
Calculations—subtraction
Calculations—moltiplication

Raw
score

Cut-off

2.09
0
0
18/18
6/6
71
20/20
9.5

5.75
5
1
–
–
52
–
8

10/10
10/10
7/7
3/3
3/3
4/4

8.8
6.3
4.2
2.2
1
1.4

Cognitive domain

Test

Raw
score

Equivalent
score

Praxis abilities

Rey-Osterrieth Complex Figure—copy (Caffarra, Vezzadini, Dieci, Zonato, & Venneri,
2002)
Weigl Sorting Test (Laiacona, Inzaghi, De Tanti, & Capitani, 2000)
Standard Progressive Matrices (Spinnler & Tognoni, 1987)
Phonemic verbal fluency (Spinnler & Tognoni, 1987)
Semantic verbal fluency (Spinnler & Tognoni, 1987)
Numerical span (Monaco, Costa, Caltagirone, & Carlesimo, 2013)
Reverse numerical span (Monaco et al., 2013)
List of 15 words (Carlesimo, Caltagirone, Fadda, Marfia, & Gainotti, 1995)

30

0

12
29/48
40
42
6
5
67/75

3
1
4
3
3
4
4

Nonverbal executive functions
Abstract reasoning
Lexical access
Verbal short-term memory
Verbal working memory
Verbal long-term memory—Immediate
recall

Prose memory (Spinnler & Tognoni, 1987)
46.5
4
Verbal long-term memory—Delayed recall List of 15 words (Carlesimo et al., 1995)
15/15
4
7
0
Visuo-spatial long-term memory
Rey-Osterrieth Complex Figure—recall (Caffarra et al., 2002)
Attention
Attentive matrices (Spinnler & Tognoni, 1987)
48
2
Raw scores and equivalent or cut-off scores are provided. For the equivalent score: 0 = pathological performance; 1 = borderline performance; 2−4 = performance
within the normal range (see Capitani & Laiacona, 1997 for psychometric details).
Abbreviations: E.N.P.A = Esame neuropsicologico per l’afasia (Capasso & Miceli, 2001).

6

C. MONTI ET AL.

Table 3. Low-level vision and object processing evaluation: results.
Cognitive domain
Low-level visual abilities

Perceptual processing and space perception

Verbal naming

Test
L-POST
Fine shape discrimination
Dot lattices
RFP contour integration
Shape ratio discrimination (Efron)
Figure ground segmentation
Kinetic object segmentation
RFP texture surfaces
RFP fragmented outline
Biological motion
Recognition of missing part
Global Motion Detection
Dot counting
Recognition of objects in isolation
Embedded figure detection
Recognition of object in a scene
Total score
Visuospatial judgment (Benton et al., 1978)
VOSP
Screening test
Incomplete letters
Silhouettes
Object decision
Progressive silhouettes
Dot counting
Position discrimination
Number location
Cube analysis
BORB
Lenght match (version A)
Lenght match (version B)
Size match (version A)
Size match (version B)
Orientation match (version A)
Orientation match (version B)
Figure naming (AAT)

Cognitive domain
Visual-perceptual analyses

Colour processing

Test
Overlapping Figures (Della Sala et al., 1995)
Concrete
Abstract
Total score
Ishihara Test for Colour Blindness
Numbers
Paths
Farnsworth-Munsell 28 HueColor Vision test (Farnsworth, 1943)
Colour naming (AAT subtest)
Colour semantic knowledge

Raw score

Cut-off

5/5
5/5
5/5
5/5
5/5
5/5
4/5
4/5*
5/5
5/5
5/5
2/5*
5/5
5/5
5/5
70/75
26

19

20
20
16
18
15
9
20
6*
7

15
17
16
15
14
8
18
7
6

24/30
21/30
23/30
27/30
25/30
24/30
40/40

23,7

Raw score

4
4
4
4
4
5
4
5
3
4
4
4
5
3
5

22,5
19,6

Equivalent score

32/36
15/35
47/71

1
0
0

21/25
9/13

17
normal

7/7
6/6

Raw scores and equivalent or cut-off scores are provided. For the equivalent score: 0 = pathological performance; 1 = borderline performance; 2–4 = performance
within the normal range (see Capitani & Laiacona, 1997 for psychometric details).
Abbreviations: L-POST = Leuven Perceptual Organization Screening Test; RFP = Radial frequency pattern; VOSP = Visual Object and Space Perception battery;
BORB = Birmingham Object Recognition Battery; AAT = Aachener Aphasie Test.

below the cut-off in the Dot counting and Radial frequency pattern (RFP) fragmented outline subtests.
Furthermore, the Overlapping Figures Test (Della
Sala, Laiacona, Trivelli, & Spinnler, 1995) was used to
examine visual-perceptual analyses. ST’s score was
below the normal range due to selective difficulties
in meaningless patterns. Nevertheless, at the Visual
Object and Space Perception battery (VOSP, Warrington & James, 1991) the patient performed well
within the typical range in all subtests evaluating perceptual processing and object perception. Regarding
space perception, ST’s scores were in the normal

range except for some difficulties in the number
location task, likely to be attributed to the presence
of hemianopia. Adequate visuospatial analysis was
confirmed from the results at the Birmingham
Object Recognition Battery (BORB, Riddoch & Humphreys, 1993) length match, size match and orientation match subtests, together with the Visuospatial
Judgment test score (Benton, Varney, & deS
Hamsher, 1978). Moreover, verbal naming was preserved (see Table 3 for a summary of the results).
Taken together, ST’s neuropsychological performance indicates intact abilities in perceptual

COGNITIVE NEUROPSYCHOLOGY

organization and space perception, except for some
difficulties in abstract shapes discrimination and isolated difficulties in dot counting, likely to be attributed
to left hemianopia. Results from perceptual and
naming tasks suggest that structural encoding and
recognition of visually presented stimuli are unimpaired. ST does not show any impairment in basiclevel object recognition. Moreover, a tactile recognition task revealed no impairments in recognizing
objects either through non-visual modality (accuracy
10/10).
2.2.6. Conclusions
A formal evaluation of ST’ face recognition abilities
strongly indicates the presence of a severe impairment in the (i) recognition of previously known personally familiar faces, (ii) ability to learn and
remember new faces (CFMT), (iii) ability to match unfamiliar faces (GFMT). Besides the presence of visual
field reduction (left homonymous hemianopia), ST
did not show visual-perceptual defects, achromatopsia, objects agnosia, and topographical disorientation,
which are known to be frequently associated in
patients with face recognition difficulties (Barton,
Press, Keenan, & O’Connor, 2002; Bouvier & Engel,
2006; Clarke et al., 1997; Gauthier, Behrmann, & Tarr,
1999). Overall, ST’s face-selective impairment,
together with the results from the neuropsychological
assessment and clinical observations indicate a diagnosis of a “pure” acquired prosopagnosia.
2.3. Holistic processing measures
2.3.1. Identity composite experiment
2.3.1.1. Participants. Eleven healthy subjects were
tested (6 females; mean age: 41 ± 13).

2.3.1.2. Materials and methods. Stimuli were
selected from the Ekman 60 Faces Test (Young et al.,
2002). Images of eight individuals were chosen (4
males; 4 females). All the faces displayed the same
expression (i.e., happiness). Images were separated
into top and bottom halves by cutting stimuli horizontally at the bridge of the nose; the two halves were
separated by a 5 pixel gap (see Ramon & Rossion,
2012, for a similar procedure). By combining top and
bottom parts we obtained a set of 32 composite
stimuli. Besides eight individuals originally selected,
which constituted congruent stimuli (i.e., top and

7

bottom parts with same identity), twenty-four were
incongruent stimuli (i.e., top and bottom parts of
different individuals with the same gender). The
same set of aligned stimuli was then arranged to
create a misaligned one. Misaligned stimuli were
formed by horizontally shifting the bottom part to
the left or to the right of the top one by about half
of the image width. Thus, 64 stimuli were obtained
(32 aligned; 16 with left bottom misalignment; 16
with right bottom misalignment). On each trial, two
face stimuli (S1 – S2) were shown sequentially. S2
was maintained congruent (top and bottom belonged
to the same identity) while we randomly varied the
congruency of S1 (congruent S1: top and bottom of
the same identity; not congruent S1: top and bottom
of different identities). Thus, we obtained “same” and
“different” conditions. For the “same” condition,
according to the congruency of S1, the trial interval
randomly showed: (i) same top halves/same bottom
halves (congruent S1) or (ii) same top halves/
different bottom halves (not congruent S1). Experimental variables were aligned/misaligned; same/
different response; S1 congruent/ not congruent;
response to the top/bottom halves (see Figure 2 for
an example of stimuli).

2.3.1.3. Procedure. The experiment was run by using
E-prime 2.0. Participants were seated at approximately
43 cm viewing distance from the monitor. Stimuli
were presented on a white background. On each
trial two composite faces (S1–S2) were sequentially
displayed and participants were requested to judge
whether either the top or the bottom halves of the
two stimuli showed the same or different identities,
by pressing the corresponding keys. Each trial
showed two aligned or misaligned composite faces.
The side of misalignment (left; right) was randomized
between trials. Each aligned stimulus was 10.8
(width) × 15.2 cm (height) (13.1 × 19.4° in visual
angle, viewed from a distance of 43 cm). Each misaligned stimulus was 16.1 (width) × 15 cm (height)
(21.4 × 19.4° in visual angle, viewed from a distance
of 43 cm). Reaction times and accuracy were recorded.
At the beginning of each trial a fixation cross was presented for 1000 ms at the centre of the screen, followed by S1 (150 ms), a second fixation cross
(150 ms) and S2 (150 ms). Participants had a
maximum of 3500 ms for response after S2 disappearance. Trials were separated by 1000 ms interstimulus

8

C. MONTI ET AL.

Figure 2. Identity composite experiment: example of stimuli. The figure shows an example of aligned (a) and misaligned (b) “same” trial,
top target condition. The participants have to judge if the top halves of two sequentially presented faces (prime and target; left and
right in the Figure) belong to the same or to different individuals.

interval (ISI) (Figure 3). Participants completed four
blocks of 64 trials differing with respect to the part
they have to classify, either the top or the bottom
target. Blocks order was fixed for each participant
(bottom, top, bottom, top). Aligned and misaligned
trials were randomized within blocks.

2.3.1.4. Analyses. In line with Rossion (2013), analyses
were conducted only considering the “same” condition. Since the trial “congruent S1—congruent S2”
with same top halves actually showed two identical
faces, they were not considered in the analyses. Therefore, analyses were conducted on critical condition’s
trials (i.e., incongruent S1—same response trials). We
performed a group-level analysis using mixed effects
models. Mixed effects models allow accounting for
both inter-subject and intra-subject variability, increasing the reliability of the model (Bates, 2015). These
analyses were performed on a single trial basis,
for both experiments (identity and emotional
expressions)
separately,
comparing
controls’

performance to ST’s (both in the initial testing and in
the 18-months follow-up).
The analyses were conducted by using generalized
mixed effects models (on a binary distribution) on accuracy (coded as 0–1), and linear mixed effects models on
response times (RTs). All models included alignment (2
levels: aligned vs. misaligned), top–bottom (TB) (2
levels: target in top vs. bottom of the face), and condition
(2 levels: controls vs. ST) as independent fixed factors (in
a full factorial design), and random intercepts. In models
on accuracy, p-values were obtained by “models comparison method” (i.e., by comparing the deviance in
two models—with vs. without the effect of interest—
in a chi-squared test). In the model on RTs, only
correct trials (accuracy = 1) were considered. Outliers
exceeding 2 SD from the mean were excluded (mean
and SD were computed on each subject). P-values, in
all post-hoc multiple comparisons, were adjusted
according to the Tukey HSD method. Only main
effects and interaction involving the condition factor
(main variable of interest) are reported below.

COGNITIVE NEUROPSYCHOLOGY

9

Figure 3. Identity composite experiment: trial procedure. Here is shown an example of same-top target condition: two same top are
paired with different bottom halves (aligned, S1 not congruent).

Moreover, we ran single-subject analyses by comparing aligned and misaligned trials for each control
subject and ST (Single-subject analyses). We here
used linear models (LM) for RTs and generalized linear
models (GLM) on binomial distribution for accuracy.
Alignment and top/bottom target were considered
as independent within-subjects variables. RTs associated with correct responses were cleaned at the individual-subject level by removing trials above or below
2 SD from the subject’s mean.

2.3.1.5. Results
2.3.1.5.1. Single-subject analyses. Accuracy: in 9 out
of 11 subjects the numerical difference is in the
direction expected for healthy participants (i.e.,
they are more accurate in misaligned rather than
aligned trials); of these, 6 show a statistically significant misaligned vs. aligned difference (Figure 4b).
RTs: in 10/11 subjects the numerical difference
between conditions is in the expected direction
(i.e., they are faster in misaligned than aligned
trials) and 2 of these show a statistically significant
misaligned vs. aligned difference (Figure 4c). ST,

instead, showed a “reversed” CFE (i.e., she is faster
and more accurate in aligned, rather than misaligned, trials), which is statistically significant in
terms of both accuracy (χ2 (1) = 9.803, p = .002)
and RTs (F (1,48) = 8.599, p = .005). See Figure 4a
for a comparison between controls’ average scores
(accuracy and RTs) and ST performance in aligned
and misaligned conditions.
2.3.1.5.2. Group-level analyses. Accuracy: the model
showed a statistically significant main effect of condition: χ 2(1) = 5.658, p = .017; and a significant alignment * condition interaction (χ 2(1) = 36.644, p < .001).
The main effect of condition showed that, on
average, control participants (m = 0.88) were more
accurate than ST (m = 0.57) in the task. In the interaction effect, post-hoc comparisons based on the
alignment showed that ST was more accurate in
aligned (m = 0.73) than misaligned (m = 0.42; z =
3.104, p = .002) trials, while healthy participants were
more accurate in misaligned (m = 0.95) than aligned
(m = 0.82; z = −6.099, p < .001) trials (i.e., ST thus
shows an opposite effect of alignment) (Figure 5a).
Comparisons based on the condition showed that ST

10

C. MONTI ET AL.

Figure 4. Identity composite experiment: single-subject analyses results. a: average observed accuracy scores (left column) and average
observed RTs (right column) in the aligned and misaligned condition for both ST and controls. b: the figure shows accuracy face composite (CFE) indexes (resulting from comparison between the aligned and misaligned condition) for each participants alongside with ST.
c: the figure shows RTs face composite indexes (resulting from comparison between the aligned and misaligned condition) for each
participants alongside with ST. Asterisks indicates significant differences between aligned and misaligned trials at the single-subject
level. Bars represent standard errors.

presented significantly lower accuracy than healthy
participants (z = −4.497, p < .001) in misaligned trials,
while, for aligned trials, ST and healthy participants
did not show any difference (z = −0.778, p = .431)
(Figure 5a).
RTs: the model showed a significant interaction
effect of alignment * condition (F(1, 918.83) = 13.085,
p < .001) (Figure 5b). Post-hoc comparisons showed
that while controls were slower in aligned trials (m =
628 ms) than in misaligned ones (m = 599 ms) (t
(924) = 3.155, p = .002), ST was slower in misaligned
(m = 881 ms) than aligned (m = 679 ms) trials (t(924)
= −3.002, p = .003). Comparisons based on the condition showed that ST presented a trend towards statistical significance for slower RTs compared to healthy
participants (t(13) = 2.126, p = .054) in misaligned
trials, while, for aligned trials, ST and healthy

participants did not show any difference (t(11) =
0.443, p = .666) (Table 4 for results).
2.3.2. Emotional expression composite experiment
2.3.2.1. Participants. Thirteen healthy subjects were
tested (12 females; mean age: 39 ± 13).

2.3.2.2. Materials and methods. Stimuli were selected
from the Ekman & Friesen 60 Faces Test (Young et al.,
2002). We chose photographs depicting four different
individuals (two males and two females), each one
showing three distinct facial expressions. According to
Baudouin and Humphreys (2006) we selected
expression of happiness, anger and fear since happiness and anger are easy to be discriminated both
from the top and the bottom part (see Baudouin &
Humphreys, 2006, for a comprehensive description).

COGNITIVE NEUROPSYCHOLOGY

11

Figure 5. Identity composite experiment: group-level analyses results. In this and following plots: dots represent the mean predicted value
and error bars represent 95% confidence intervals; the blue line represents ST’s scores, while the red line represents average control
participants’ scores. Panel A: Plot representing the alignment * condition interaction effect on participants’ accuracy. Panel B: Plot representing the same effect on participants’ reaction times.

Moreover, such expressions are well recognized by our
patient (see results from Ekman & Friesen test).
For each of the four individuals selected, all possible
combinations were created, obtaining 36 composite

Table 4. ST’s and control’s performance at the Identity
composite (Experiment 1) and Emotional expression composite
(Experiment 2) tasks.
a. Top target.
Testing
Identity composite
Accuracy
ST
Controls
Aligned
62,5
76.1 (42.7)
Misaligned
58,3
97.7 (14.9)
Emotional expression composite
Accuracy
ST
Controls
Aligned
75,0
76.0 (42.8)
Misaligned
41,7
92.3 (26.7)

Follow up
RTs
ST
836
875

Controls
672 (238)
620 (233)

Accuracy
ST
50
70,8

RTs
ST
624
686

RTs
ST
803
864

Controls
790 (330)
709 (317)

Accuracy
ST
75,0
66,7

RTs
ST
623
625

RTs
ST
569
892

Controls
593 (223)
578 (202)

Accuracy
ST
87,5
54,2

RTs
ST
549
667

RTs
ST
672
835

Controls
722 (360)
749 (296)

Accuracy
ST
91,7
75,0

RTs
ST
583
731

b. Bottom target.
Identity composite
Accuracy
ST
Controls
Aligned
83,3
87.5 (33.1)
Misaligned
25,0
92 (27.1)
Emotional expression composite
Accuracy
ST
Controls
Aligned
95,8
80.4 (39.7)
Misaligned
58,3
92.0 (27.2)

Means and standard deviations (in brackets) are showed. Left column: testing
phase; Right column: follow-up evaluation. The Table shows results from
the top-target (Table 4a) and bottom-target condition (Table 4b).

stimuli (12 congruent, 24 incongruent). Each incongruent composite depicted a face constituted by two
halves of the same individual showing different
emotions: one on the top half and a different one on
the bottom half (e.g., happiness on the top and fear
on the bottom). Congruent composites were made by
two halves showing a unique emotion expressed by
the same individual. A correspondent misaligned set
of stimuli was created through the same procedure as
in Experiment 1. See Figure 6 for an example of stimuli.

2.3.2.3. Procedure. Experimental procedure was the
same as in Experiment 1. To avoid possible perceptual
interference generated from identity information, in
each trial faces of the same individual were shown,
which could be congruent or incongruent for
emotional expression. Subjects were required to
judge whether the two upper or lower halves of the
faces showed the same or different emotions. We
stressed that they did not have to judge the identity
of the faces; rather they had to attend to the
emotional expression by ignoring the identity. Participants completed four blocks of 80 trials (response to
the top; bottom; top; bottom).
2.3.2.4. Analyses. Analyses were the same as for
Experiment 1.

12
C. MONTI ET AL.

Figure 6. Emotional expression composite experiment: example of stimuli. A matrix example of stimuli used. Each pair displays the aligned and the correspondent misaligned version. Here are
shown stimuli for the congruent (upper row) and not congruent (second and third row) conditions. Congruent stimuli depicted a face showing a unique facial expression of the same
identity (from left to right: happiness; fear; anger). Not congruent stimuli are composite faces in which the upper and lower halves have same identity but differ for the expressed emotions.
From left to right, second row: top happiness-bottom fear; top anger-bottom happiness; top anger-bottom happiness. From left to right, third row: top happiness-bottom anger; top fearbottom anger; top anger-bottom fear. Four different face identities were used.

COGNITIVE NEUROPSYCHOLOGY

2.3.2.5. Results
2.3.2.5.1. Single-subject analyses. Accuracy: in 12 out of
13 subjects the numerical difference is in the expected
direction (“typical” CFE); 8 out of 12 show a statistically
significant misaligned vs. aligned difference. ST
showed a statistically significant reversed CFE (χ2 (1)
= 15.033, p = .001) (Figure 7b). See Figure 7a for a comparison between controls’ average scores (accuracy
and RTs) and ST performance in aligned and misaligned conditions.
RTs: in 8/13 subjects there is a typical CFE (2 of these
show a statistically significant misaligned vs. aligned
difference). ST showed a reversed CFE in a trend
toward significance (F (1,57) = 3.435, p = .07) To note,

13

that also two controls displayed a statistically significant “reversed” CFE (Figure 7c). However, when
looking at both accuracy and RTs, we note that
these subjects showed a trade-off between accuracy
and correct RTs (i.e., they were faster in aligned trials,
but they were better in misaligned trials; subject N4,
RTs: aligned 518ms, misaligned 647ms; accuracy:
aligned 0.83, misaligned 1. Subject N5, RTs: aligned
455ms, misaligned 546 ms; accuracy: aligned 0.83,
misaligned 0.85). It is also important to note that the
CFEs for RTs and accuracy have opposite directions
for subjects N4 and N5; whereas this is not evident
for ST, who showed a congruent reversed-CFE for
both accuracy and RTs (Figure 7b and c).

Figure 7. Emotional expression composite experiment: single-subject analyses results. a: average accuracy scores (left column) and
average RTs (right column) in the aligned and misaligned condition for both ST and controls. b: the figure shows accuracy face composite (CFE) indexes (resulting from comparison between the aligned and misaligned condition) for each participants alongside with ST.
c: the figure shows RTs face composite indexes (resulting from comparison between the aligned and misaligned condition) for each
participants alongside with ST. Asterisks indicates significant differences between aligned and misaligned trials at the single-subject
level. Bars represent standard errors.

14

C. MONTI ET AL.

To take into account the above-mentioned tradeoffs, we have computed a global measure of CFE
that combines accuracy rates and correct RTs (see
Busigny & Rossion, 2010, for a similar procedure
applied to the face inversion effect in a case of AP)
and we have compared ST with the control’s mean
using SINGLIMS t-test for small samples (Crawford
et al., 2010). To do this, we calculated the Inverse
Efficiency score (IE) for each participant and for ST
(IE = RTs/accuracy) then we calculated the CFE index
using the following formula: (IE aligned—IE misaligned)/(IE aligned + IE misaligned). Our results
showed that ST has a significantly lower CFE index
(−0.33) in comparison to the control participants
(mean: 0.1; t(12) = −265.8, p < .001, two-tailed).
2.3.2.5.2. Group-level analyses. Accuracy: the model
showed a statistically significant alignment * condition
interaction: χ 2(1) = 38.3, p < .001 (Figure 8a). Post-hoc
comparisons based on alignment showed that ST
was more accurate in aligned (m = 0.85) than misaligned (m = 0.50; z = 3.361, p < .001) trials, while
healthy participants were more accurate in misaligned
(m = 0.92) than aligned (m = 0.78; z = −6.780, p < .001)
trials. Comparisons based on the condition showed
that ST presented significant lower accuracy than
healthy participants (z = −3.916, p < .001) in misaligned trials, while, for aligned trials, ST and healthy
participants did not show any difference (z = −0.944,
p = .345).

RTs: The model on participants’ RTs showed a significant alignment * condition interaction (F(1,
1043.28) = 4.142, p = .042) (Figure 8b). Post-hoc comparisons based on the alignment showed that ST did
not present a statistically significant difference
between aligned (m = 729 ms) and misaligned trials
(m = 848 ms) (t(1046) = −1.640, p = .101), while
healthy participants showed a trend towards shorter
RTs in misaligned (m = 729 ms) than aligned trials
(m = 755 ms) (t(1046) = 1.904, p = .057). Comparisons
based on the condition did not highlight any statistical
significant difference between ST and healthy participants, neither in aligned (t(12) = −0.137, p = .893) or in
misaligned trials (t(13) = 0.493, p = .630) (Table 4 for
results).
2.3.3. Control analyses
ST’s neuropsychological assessment highlighted left
hemianopsia. It is thus possible that ST’s problems
shown in the CFE (identity and emotions) are due to
a low-level deficit rather than holistic processing
impairments. This is unlikely since the offset of the
misaligned face was equally distributed between left
and right. However, to exclude potential low-level
effects on ST’s performance we ran additional eight
control analyses focusing on the effect of trials N−1.
In both Experiments, aligned and misaligned trials
were interleaved. Thus, on trial N participants may
fixate on the top (or bottom) half of an aligned face.

Figure 8. Emotional expression composite experiment: group-level analyses results. Panel A: Plot representing the alignment * condition
interaction effect on participants’ accuracy. Panel B: Plot representing the same effect on participants’ reaction times.

COGNITIVE NEUROPSYCHOLOGY

On trial N + 1 the misaligned face may be offset to the
left or to the right-hand side. Consequently, there is
the possibility that the patient fixated in the same
location during the two trials’ presentation and, in
the trials where the misaligned face was offset to the
left side, the task-relevant part of the face is more
likely to be in the blind visual field.
For this reason, we separately analyzed: (i) trials in
which participants view an aligned trial followed by a
misaligned trial (a condition in which the patient may
fixated in the same location), and (ii) trials in which a
misaligned trial is preceded by another misaligned
trial (in this latter case the patient might be inclined
to fixate on the location of the misaligned top face
half). We expected that, if the hemianopia is
playing a role in these results, the performance
should be worse in aligned trials that are followed
by misaligned trials than on misaligned trials followed by misaligned trials. On the other side, if ST
fixated on the central cross throughout, we do not
expect to find such a difference. Control analyses
were performed for each Experiment, for both
testing and follow up.
As for our other results (see Results section), the
analysis was conducted on top-target only since
the CFE on the bottom half is generally weaker
than the CFE on the top half (Rossion, 2013). We considered trials in which the target was aligned or misaligned to the left side. Eight models were run on
these data, considering accuracy and reaction
times (RTs) in the Identity task (test and follow up)
and in the Emotion task (testing and follow up). All
models performed on RTs were linear models,
while all models performed on the accuracy were
general linear models on a binomial distribution, in
which the statistical significance of each effect was
tested by using likelihood ratio tests (LRTs). Each
model included the following independent variables: Alignment (2 levels: Aligned vs. Misaligned
Left), N−1 Alignment (3 levels: Aligned, Misaligned
Left, Misaligned Right), and their interaction. Outliers
exceeding 2 SD from the mean were excluded from
RTs.
Results showed that none of the main effects of N
−1 Alignment or the interactions N−1 Alignment *
Alignment resulted as statistically significant. Such
findings allowed us to exclude possible effects on
the performance due to ST’s left hemianopia (see Supplementary Material for further details).

15

3. Follow up
At the follow up examination (18 months after the
acute event) ST reported persisting difficulties in familiar face recognition, which were still evident in the
clinical setting. Her prosopometamorhopsia was
almost completely recovered since she reported to
do not experience facial misperception as before.
The reduction of the left hemifield was still present
at the neuroophthalmological visit. ST was retested
with the CFE, for both identity and emotions, and
the Ekman 60 Faces Test (Young et al., 2002). Additionally, we used the face inversion effect (FIE) paradigm
(Yin, 1969) in order to examine ST with a further
measure of holistic processing.
Moreover, four statistical analyses were run to test
possible differences in ST’s performance in the Identity
and Expressions experiments between the first
measurements and follow up (Analyses pre–post).
Two generalized linear models (on a binomial distribution) were used to test ST’s accuracy in both identity
and emotional expressions tasks, while two linear
models were used to test ST’s RTs in the same tasks.
All these models included alignment, TB and time (2
levels: pre vs. post) as independent factors, in a full factorial model. RTs from incorrect trials or exceeding 2
SD from the mean were excluded from the analyses.

3.1. Identity composite experiment
3.1.1. Analyses pre–post
Accuracy: The analyses did not show any statistically significant effect involving the factor time (all χ 2 < 2.5, all p
> .1). RTs: The analyses on RTs highlighted a statistically
significant main effect of time: F(1, 103) = 11.24, p
= .001 (identity task) (Figure 9). RTs in the first measurement (mean = 749 ms) were slower than in the follow
up experiment (mean = 622 ms). No other interaction
effects including the factor time resulted as statistically
significant (all Fs < 2.1, all ps > .15).
3.1.2. Group-level analyses
Accuracy: The model on participants’ accuracy showed
a statistically significant main effect of condition: χ 2(1)
= 4.0273, p = .045; and a significant alignment * condition interaction: χ 2(1) = 15.235, p < .001 (Figure
10a). The main effect of condition showed that, on
average, control participants (m = 0.88) were more
accurate than ST (m = 0.66) in the task. In the

16

C. MONTI ET AL.

Figure 9. Identity composite experiment: Boxplot representing ST’s RTs during the first measurement (after the acute phase) and during
the follow-up, in the identity composite experiment. This plot highlights the statistically significant main effect of the factor Time in the
linear model.

interaction effect, post-hoc comparisons based on the
alignment indicated that ST did not show a statistically
significant difference between aligned (m = 0.69) and
misaligned (m = 0.63; z = 0.934, p = .350) trials, while
healthy participants were more accurate in misaligned
(m = 0.95) than aligned (m = 0.82; z = −6.099, p < .001)
trials. Comparisons based on the condition led to the
same pattern found in the first measurement: ST presented lower accuracy than healthy participants in
misaligned trials (z = −3.402, p < .001), while for
aligned trials ST and healthy participants did not
differ (z = −0.886, p = .375).
RTs: the model on participants’ RTs showed a significant alignment * condition interaction (F(1,
929.05) = 6.0597, p = .014) (Figure 10b). Post-hoc
comparisons based on the alignment showed that
ST presented an opposite trend effect direction compared to healthy participants: while healthy participants were slower in aligned trials (m = 628 ms)
than in misaligned ones (m = 599 ms) (t(931) =
3.207, p = .001), ST was slower in misaligned trials
(m = 678 ms) than aligned ones (m = 575 ms) (t
(931) = −1.727, p = .085). Comparisons based on the
condition showed no differences between ST and
healthy participants, neither in misaligned (t(11) =
0.572, p = .579) or in aligned trials (t(11) = −0.470,
p = .648). See Table 4 for results.

3.2. Emotional expression composite experiment
3.2.1. Analyses pre–post
Accuracy: The analyses did not show any statistically
significant effect involving the factor time (all χ 2 <
2.5, all p > .1). RTs: The analyses on RTs highlighted
a statistically significant main effect of time: F(1,
123) = 14.60, p < .001 (Figure 11). RTs in the first
measurement (mean = 772 ms) were slower than
in the follow up experiment (mean = 637 ms). No
other interaction effects including the factor time
resulted as statistically significant (all Fs < 2.1, all
ps > .15).
3.2.2. Group-level analyses
Accuracy: The model on participants’ accuracy showed
a significant alignment * condition interaction (χ 2(1) =
14.213, p < .001) (Figure 10c). Post-hoc comparisons
based on the alignment showed the same pattern
found in the identity task: ST did not present a significant difference between aligned (m = 0.83) and misaligned (m = 0.71; z = 1.581, p = .114) trials, while
healthy participants were more accurate in misaligned
(m = 0.92) than aligned (m = 0.78; z = −6.780, p < .001)
trials. Comparisons based on the condition led to the
same pattern found in the first measurement: ST presented significantly lower accuracy than healthy

COGNITIVE NEUROPSYCHOLOGY

17

Figure 10. Follow up experiments: Panel A and B: Plots representing the alignment * condition interaction effect on participants’ Accuracy (A) and reaction times (B) in the follow up identity composite experiment. Panel C and D: Plots representing the same effect on
participants’ Accuracy (C) and reaction times (D) in the follow up emotional expression composite experiment.

participants (z = −2.496, p = .013) in misaligned trials,
while, for aligned trials, ST and healthy participants
did not show any statistically significant differences
(z = 0.543, p = .587).
RTs: the model on participants’ RTs showed a trend
towards significance in the alignment * condition interaction (F(1, 1055.98) = 2.809, p = .094) (Figure 10d).
Even though this effect is not strictly statistically significant, multiple comparisons were performed to
compare this effect with the same effect in the first
measurement. Post-hoc multiple comparisons on the

interaction effect did not lead to clear differences, still
reflecting the same pattern found in the first measurement. Comparisons based on the alignment showed
that ST did not present a statistically significant difference between aligned (m = 601 ms) and misaligned
trials (m = 683 ms; t(1053) = −1.226, p = .220), while
healthy participants showed a trend towards shorter
RTs in misaligned (m = 729 ms) than aligned trials (m
= 755 ms) (t(1053) = 1.927, p = .054). Comparisons
based on the condition did not highlight any significant
difference between ST and healthy participants, neither

18

C. MONTI ET AL.

Figure 11. Emotional expression composite experiment: Boxplot representing ST’s RTs during the first measurement and during the
follow-up, in the emotional expression composite experiment. This plot highlights the statistically significant main effect of the
factor Time in the linear model.

in aligned (t(12) = −0.737, p = .475) or in misaligned
trials (t(13) = 0.265, p = .795) (Table 4 for results).
3.3. Ekman 60 faces Test
ST’ s expression recognition ability was at normal
range (raw score: 48/60; corrected score: 46.05/60,
adjusted for age, gender and education; equivalent
score: 3; Dodich et al., 2014).

3.4. Face inversion experiment
3.4.1. Material and methods
Stimuli were selected from the Face Database
(Minear & Park, 2004). Thirty-two stimuli were
chosen (16 females; 16 males) and divided in four
age range subsets (18–29; 30–49; 50–69; 70–94
years old). All stimuli were modified in greyscale
and cropped to remove extra-face stimuli. For each
trial a fixation cross was presented at the centre of
the screen (1000 ms), followed by the target (3000
ms), a fixation cross (2000 ms) and a slide with
two simultaneously presented probes (3000 ms). Participants were instructed to select one of the two
faces (left or right) matching the one previously
shown, by pressing the corresponding key. The
experiment was divided in four blocks of 32 trials

differing for stimuli orientation, for a total of 128
trials. Blocks order was fixed for each participant
(upright, inverted, upright, inverted). The experiment
was run with E-Prime 2.0. A face inversion score was
calculated as following: inverted-upright for correct
RTs, and upright-inverted for accuracy. ST’s performance was compared to that of five healthy subjects
(3 females; mean age: 44 ± 13) by means of a t-test
for small samples (SINGLIMS t-test; Crawford et al.,
2010).
3.4.2. Results
ST’s face inversion score for RTs was significantly lower
as compared to the controls’ average score (ST: 20.2;
control’s mean: 78.7, SD = 32.7; t(4)= −2,761, p = .03,
one tailed). ST’s face inversion score for accuracy did
not differ from the control’s (ST: 3.1; control’s mean:
2,2, SD = 4,1; t(4)= .20, p = .42, one-tailed). Thus,
although accurate, ST failed to show an upright superiority for RTs, as opposite to control subjects. It is
known that correct RTs are a highly sensitive
measure of the effects of face inversion in normal
observers (Rossion, 2008).
To conclude, severe difficulties in face recognition
were still present at the follow up evaluation. As for
the initial assessment, ST showed reduced identityCFE. The absence of face inversion effect further

COGNITIVE NEUROPSYCHOLOGY

confirmed the presence of atypical holistic processing
in our patient. Moreover, she still showed atypical
expressions-CFE despite spared emotion recognition
ability. As compared to the initial assessment, a
decrease in ST’ reaction times was observed for both
Identity and Expressions experiments, congruent
with a general improvement which could be ascribed
to a spontaneous recovery or to the adoption of perceptual strategies.

4. General discussion
In the present study we examined the sensitivity to the
CFE for both identity and facial expressions in a case of
AP (ST) who was severely impaired in identity, albeit
not in emotion recognition. First, we assessed
whether the impairment in face recognition was
associated with the lack of perceptual integration of
face parts into a whole. To this aim, we examined
ST’s sensitivity to the CFE for identity processing
(Experiment 1). Second, we tested the expressionsCFE to ascertain whether her spared facial expressions
recognition relies on holistic mechanisms or, rather, on
“compensatory” part-based strategies (Experiment 2).
Results from Experiment 1 (Identity composite)
showed a “typical” composite face effect in the
control group, characterized by better performance
for misaligned than aligned trials. The difference in
the performance between matching two same top
halves when their different bottom halves are
aligned Vs. misaligned is considered as an index of
perceptual integration of the irrelevant part with the
attended one (Rossion, 2013). This (typical) integration, where misaligned trials are better and faster
processed than aligned, is seen in controls only. ST,
on the other hand, showed an opposite pattern (i.e.,
better performance on aligned trials), thus suggesting
atypical holistic coding. The reduced face-inversion
effect for ST emerged in the follow-up examination
provided further support to such interpretation. Congruently, performance on the GFMT (Burton et al.,
2010) revealed severe difficulties when ST had to
match faces varying for within-identity changes,
which might be ascribed to a low-level perceptual
difficulty characterized by weak holistic integration
of face features. Our results are in line with evidences
of weak or atypical CFE in APs (Busigny et al., 2010;
Ramon et al., 2010). However, findings indicating
typical sensitivity to the CFE in APs (see Finzi et al.,

19

2016; Rezlescu et al., 2012) and in CPs (Biotti et al.,
2017), pointed out some variability in these populations which can be due to different experimental
manipulations or, in the case of APs, to different
location/size of brain damage.
The presence of an atypical CFE in a case of AP is
consistent with the proposal that holistic processing
contributes to efficient face recognition. The link
between holistic coding and face recognition ability
was demonstrated in a study of DeGutis et al. (2013):
using a regression-based approach, authors found
that both part-whole and composite measures
showed robust correlations with the CFMT. As
authors suggested, it should be noted that holistic
processing is not the unique contributor to efficient
face recognition, since holistic face processing and
non-holistic processing were found to independently
predict face recognition ability. Furthermore, in
support to the role of holistic coding in face recognition, results from Richler, Cheung, and Gauthier
(2011) indicated that holistic processing predicted
face recognition abilities on the CFMT and in a task
of face identification.
The second aim of the current study was to
examine whether typical facial expressions recognition in our patient relies on holistic mechanisms or,
rather, it might be supported by “part-based” strategies. Results from Experiment 2 (Facial expression
composite) are similar to those from Experiment 1
(Identity expression composite): the control group
showed a misalignment advantage (i.e., a typical
CFE), whereas ST had an opposite pattern (i.e., better
performance on aligned trials).
Taken together, results from the two composite
tasks demonstrate that holistic processing difficulties
for both identity and facial expressions can be
involved in a patient with AP, and that these difficulties cannot be explained by low-level (i.e., left hemianopsia) problems (see Supplementary Material). To the
best of our knowledge, there are only few studies
examining CFE for facial expressions in APs. Interestingly, one study examining the sensitivity to the composite interference in a patient with AP (HJA) showed
a residual ability to judge facial emotions although it
was qualitatively different with respect to the controls
(Baudouin & Humphreys, 2006). Specifically, HJA did
not show a configural interference effect for facial
emotion judgements, leading authors to suppose
that he relied on the analyses of critical local features.

20

C. MONTI ET AL.

Nevertheless, the task involved high-order categorization abilities given that it required naming emotions.
Thus, to date, little is known on how facial expressions
are processed at a low-level of perceptual processing
in patients with face recognition impairments. Furthermore, even though HJA seemed not to benefit from
configural information for face discrimination, he
was not formally assessed with a CFE paradigm for
identity processing.
Moreover, our results showed that facial emotions
recognition can be spared in a patient with altered
holistic coding for facial expressions. Actually, results
from the emotion recognition task (Ekman 60 faces
test) and the Facial expressions composite task indicated that ST was able to recognize facial expressions
even in the presence of atypical holistic processing for
emotions. We suppose that ST’s residual ability in
recognizing facial expressions might be ascribed to
the adoption of part-based, compensatory strategies,
based on a piecemeal analysis of face parts. Considering ST’s severe face recognition difficulties, our
findings seem to suggest that alternative, partbased, mechanisms did not support efficient identity
recognition in our case. The presence of typical
emotion recognition despite reduced expressionsCFE may also indicate the possibility that emotion recognition is not necessarily mediated by holistic processing. Congruently, studies in healthy participants
showed that expression and identity rely on different
processing styles, with expressions judgement
relying on a more piecemeal analysis of local features
(e.g., Tanaka et al., 2012). Even though some studies
provided data in favour of holistic processing for
facial expressions in healthy individuals (e.g., Calder
et al., 2000; Calder & Jansen, 2005), other studies
have demonstrated that recognition of facial
expressions involves analytic processing (Blais et al.,
2012; Bombari et al., 2013; Eisenbarth & Alpers, 2011;
Scheller, Büchel, & Gamer, 2012; Smith, Gosselin, Cottrell, & Schyns, 2005).
It is still debated the extent to which emotion and
identity processing are independent or not (e.g.,
Calder & Young, 2005). We reported a dissociation
between identity and facial expressions recognition
in a patient with AP; however, such segregation was
only evident at a high semantic level (i.e., recognition
and naming phase), thus suggesting an initial stage of
holistic integration common to both identity and
emotion processing. The fact that the two

mechanisms diverge, at least at a certain level, provides support to neuropsychological case studies
showing double dissociations (i.e., prosopagnosics
with identification deficit and preserved emotion recognition, e.g., Riddoch et al., 2008; Tranel et al., 1988
Vs. patients with impaired emotional but not identity
recognition e.g., Young, Newcombe, de Haan, Small,
& Hay, 1993), and is also in line with cognitive and
neural models on anatomical and functional distinct
systems subserving emotions and identity processing
(e.g., Bruce & Young, 1986; Haxby et al., 2000). Identity
and emotions processing dissociations were also
observed in CPs individuals, in which the facial
expressions recognition (e.g., Duchaine et al., 2003,
2007; Palermo et al., 2011) and matching (Fisher,
Towler, & Eimer, 2017) is generally unimpaired,
despite their face recognition difficulties. CPs also
showed atypical holistic processing measured with
different tasks (e.g., Avidan, Tanzer, & Behrmann,
2011; DeGutis, Cohan, Mercado, Wilmer, & Nakayama,
2012; Duchaine et al., 2007) and they were found to
adopt a part-based processing in the identity matching of external and internal features (Towler, Fisher,
& Eimer, 2018).
However, what remains controversial is whether
the dissociation between identity and emotion processing reflects early perceptual mechanisms or later
stages requiring memory and semantic-related retrieval information. From a behavioural perspective, our
findings support a model in which a common stage
of emotion and identity processing exists, at early perceptual stages. Congruently, a strong interaction
between the two mechanisms was found in healthy
individuals (Fisher et al., 2016); here, early perceptual
symmetrical interference effects between identity
and emotions have been reported (i.e., the unattended identity and the unattended expression,
which were varied orthogonally, both interfered with
the performance in the expression and the identity
matching tasks, respectively). Event-related potentials
(ERPs) recording of the N250r components revealed
that identity and expression processing interact at
an early stage, when visual representations are
matched with those stored in face memory (Fisher
et al., 2016).
Our results strictly parallel findings of CPs, with
impaired identity recognition but spared expression
recognition in the presence of weakened holistic processing of both expression and identity (Palermo et al.,

COGNITIVE NEUROPSYCHOLOGY

2011). This is consistent with a model proposing an
early, holistic coding stage for multiple facial attributes
(Model B, see Palermo et al., 2011) in which the split
between expression and identity coding occurs after
a common stage of holistic processing. Authors
suggested that CPs used atypical compensatory strategies for expressions recognition, such as an overreliance on non-holistic cues. Here we propose to
extend such model by integrating a stream of featural
processing which can interact with holistic mechanisms (Figure 12). It has, in fact, been demonstrated
that facial emotion recognition recruits dissociable,
but interactive dorsal and ventral routes in the core
face processing network (Meaux & Vuilleumier,
2016): the regions involved differ with respect to the
sensitivity to holistic, as compared to analytic facial
coding, and their engagement was modulated by
task demands. In particular, holistic information activated fusiform, inferior occipital areas and amygdala,
whereas the perception of local features mainly
engaged STS and prefrontal areas.
These findings corroborated a dual-code view in
which emotion processing may be neither strictly holistic nor featural, and that the relative contribution of
such processes varies between different expressions
(Meaux & Vuilleumier, 2016). This allows overcoming
a strictly dichotomous segregation between analytic
and holistic coding of facial information. It is plausible
that expression coding in healthy individuals is
mediated by an interaction between holistic and
part-based mechanisms and that the extent to which

Figure 12. Model of processing for identity and expressions mechanisms (adapted from Palermo et al., 2011). An early, perceptual
holistic coding is common to identity and expression processing.
Holistic processing and part-based analyses interact at an early
stage of analyses. However, whereas holistic processing seems
to be essential to an adequate identity recognition, expression
coding is likely mediated by both holistic and part-based mechanisms. When holistic processing is disrupted or altered, partbased mechanisms can help achieving expression, but not identity recognition (or in a lesser extent).

21

these processes are engaged depends on task or
expression type (Meaux & Vuilleumier, 2016).
However, by considering our findings and those
observed in CPs (Palermo et al., 2011), when holistic
processing is weak or damaged, featural processing
compensates for emotions, but not for identity recognition. Moreover, ST’s atypical holistic processing is
congruent with occipito-temporal damage, whereas
part-based mechanisms are subserved by STS and prefrontal regions (Meaux & Vuilleumier, 2016), which
were spared in our patient. Taken together, our
results are in line with other findings suggesting that
holistic processing is not mandatory for accurate
facial expressions recognition (Tanaka et al., 2012).
Specifically, it is possible that emotion recognition
relies on a continuum between featural and holistic
mechanisms (Farah, 1991) depending on stimulus
ambiguity and the type of emotion displayed
(Tanaka et al., 2012).
It has to be noted that, although difficulties in
recognizing facial expressions are described as relative
uncommon in CPs (e.g., Humphreys et al., 2007; Lee
et al., 2010; Palermo et al., 2011; Rivolta et al., 2014b;
2010), a recent study of Biotti and Cook (2016) found
CPs (particularly those individuals with apperceptive
profile) exerted difficulties in classifying facial
expressions. Authors investigated emotion recognition in CPs with the use of different measures. Participants were requested to make binary
classifications if whole-face expressions in morphed
stimuli (Experiment 1); to judge facial emotion using
only the eye-region (Experiment 2); and to classify
emotions from vocal affect (Experiment 3). CPs
showed face-specific expression recognition deficits.
Such difficulties have not been ascribed to aberrant
holistic processing given that they were evident
both when participants were asked to categorize
emotions using cues from the whole and from the
eye region in isolation. It is possible that the ability
to detect difficulties in emotion recognition may be
sensitive to the procedure used.
At this regard, a potential limitation of our study is
the use of a single measure of emotion recognition.
The Ekman 60 Faces Test is a well-known neuropsychological test assessing facial expression recognition
and it is widely used in the face processing literature
(see Mestry et al., 2012; Bate et al., 2015 in acquired
and Palermo et al., 2011 in congenital form of prosopagnosia). Identity and emotions measures used in

22

C. MONTI ET AL.

our study have long stimuli presentation durations
(Ekman 60 Faces: 5s; CFMT, GFMT and familiar face recognition: until subject’s response) so we can suggest
that they do not differ for task difficulty. However,
we cannot rule out that the adoption of more sensitive
tests and the analysis of RTs (which were not collected
in the current study) might have highlighted subtle
deficits in emotion recognition. Future studies on individuals with prosopagnosia might examine performances in different measure of facial expressions,
involving for example both static and dynamic
stimuli. A further potential limitation concerns the
use of stimuli showing facial expression of happiness
in the Identity Composite experiment. Nevertheless,
since we kept constant the emotion by varying the
identity (all faces showed happy expressions), it is unlikely that unexpected “interference” might have
affected the data. Moreover, our subjects and ST
were carefully instructed to attend solely to the identity by ignoring the emotional expressions before
running the experiment. A final point of reflection concerns the sensitivity of the CFE as a “diagnostic” tool in
clinical neuropsychology. With respect to the group of
controls, ST showed an “opposite” CFE (i.e., aligned
more accurate and faster than misaligned trials),
with clear deficits in the misaligned condition. We
exclude that this atypical performance might be due
to problems in understanding tasks instructions
since ST did not manifest any problem in understanding instructions of the (numerous) tasks adopted in
the neuropsychological assessment. Critically, even
one control on the composite identity (accuracy data
—Figure 4), and two controls on the composite
expression (RTs data—Figure 7), showed a CFE in the
direction of ST’s; these participants however, by
definition, did not show any problem with faces.
These results might indeed suggest that the composite-face task is not always valid in clinical neuropsychology, and that multiple tests should be adopted
to assess holistic processing in prosopagnosia. ST’s
atypical inversion effect (at follow-up) further corroborates her holistic deficits.
In conclusion, our findings demonstrated that holistic processing difficulties for both identity and facial
expressions can be involved in a patient with severe
AP. The atypical pattern we found for both identity
and expressions-CFE suggested that her altered holistic coding affected both components of perceptual
judgement. Such results provide further data in

support to a common stage for identity and emotion
processing at low-perceptual level. Since our patient
demonstrated difficulties for identity, but not facial
expressions recognition, our results might be indicative of a segregation of the two mechanisms at a
high semantic level, showing that facial expression
processing is achievable even with holistic coding
difficulties, possibly through compensatory, partbased mechanisms.

Disclosure statement
No potential conflict of interest was reported by the authors.

References
Avidan, G., Tanzer, M., & Behrmann, M. (2011). Impaired holistic
processing in congenital prosopagnosia. Neuropsychologia,
49(9), 2541–2552.
Barton, J. J., Press, D. Z., Keenan, J. P., & O’Connor, M. (2002).
Lesions of the fusiform face area impair perception of facial
configuration in prosopagnosia. Neurology, 58(1), 71–78.
Bate, S., Bennetts, R., Mole, J.A., Ainge, J., Gregory, N.J., Bobak,
A.K., & Bussunt, A. (2015). Rehabilitation of face-processing
skills in an adolescent with prosopagnosia: Evaluation of an
online perceptual training programme. Neuropsychological
Rehabilitation, (September), 1–30. http://doi.org/10.1080/
09602011.2014.973886
Bates, D, Mächler, M, & Bolker, B. (2015). Fitting Linear MixedEffects Models Using lme4. Journal of statistical software, 67
(1).
Baudouin, J.-Y., & Humphreys, G. W. (2006). Compensatory strategies in processing facial emotions: Evidence from prosopagnosia. Neuropsychologia, 44(8), 1361–1369. doi:10.1016/
j.neuropsychologia.2006.01.006
Benton, A. L. (1990). Facial recognition 1990. Cortex, 26,
491–499.
Benton, A. L., Varney, N. R., & deS Hamsher, K. (1978).
Visuospatial judgment: A clinical test. Archives of Neurology,
35(6), 364–367.
Biotti, F., & Cook, R. (2016). Impaired perception of facial
emotion in developmental prosopagnosia. Cortex, 81,
126–136.
Biotti, F., Wu, E., Yang, H., Jiahui, G., Duchaine, B., & Cook, R.
(2017). Normal composite face effects in developmental
prosopagnosia. Cortex, 95, 63–76. doi:10.1016/j.cortex.2017.
07.018
Blais, C., Roy, C., Fiset, D., Arguin, M., & Gosselin, F. (2012). The
eyes are not the window to basic emotions.
Neuropsychologia, 50(12), 2830–2838.
Bodamer, J. (1947). Die prosop-agnosie. Archiv für Psychiatrie
und Nervenkrankheiten, 179(1-2), 6–53.
Bombari, D., Schmid, P. C., Schmid Mast, M., Birri, S., Mast, F. W.,
& Lobmaier, J. S. (2013). Emotion recognition: The role of

COGNITIVE NEUROPSYCHOLOGY

featural and configural face information. The Quarterly
Journal of Experimental Psychology, 66(12), 2426–2442.
Bouvier, S. E., & Engel, S. A. (2006). Behavioral deficits and cortical damage loci in cerebral achromatopsia. Cerebral
Cortex, 16(2), 183–191.
Bowles, D. C., McKone, E., Dawel, A., Duchaine, B., Palermo, R.,
Schmalzl, L., … Yovel, G. (2009). Diagnosing prosopagnosia:
Effects of ageing, sex, and participant–stimulus ethnic
match on the Cambridge Face Memory Test and Cambridge
Face Perception Test. Cognitive Neuropsychology, 26(5), 423–
455.
Bruce, V., & Young, A. (1986). Understanding face recognition.
British Journal of Psychology, 77, 305–327. doi:10.1111/j.
2044-8295.1986.tb02199.x
Burton, A.M., White, D., & McNeil, A. (2010). The Glasgow Face
Matching Test. Behavior Research Method, 42, 286–291.
doi:10.3758/BRM.42.1.286
Burton, A. M., White, D., & McNeill, A. (2010). The Glasgow face
matching test. Behavior Research Methods, 42(1), 286–291.
Busigny, T., Joubert, S., Felician, O., Ceccaldi, M., & Rossion, B.
(2010). Holistic perception of the individual face is specific
and necessary: Evidence from an extensive case study of
acquired prosopagnosia. Neuropsychologia, 48(14), 4057–
4092.
Busigny, T., & Rossion, B. (2010). Acquired prosopagnosia
abolishes the face inversion effect. Cortex; a Journal
Devoted to the Study of the Nervous System and Behavior,
46(8), 965–981.
Busigny, T., Van Belle, G., Jemel, B., Hosein, A., Joubert, S., &
Rossion, B. (2014). Face-specific impairment in holistic perception following focal lesion of the right anterior temporal
lobe. Neuropsychologia, 56, 312–333.
Caffarra, P., Vezzadini, G., Dieci, F., Zonato, F., & Venneri, A.
(2002). Reyeosterrieth complex figure: Normative values in
an Italian population sample. Neurological Sciences, 22(6),
443–447.
Calder, A. J., & Jansen, J. (2005). Configural coding of facial
expressions: The impact of inversion and photographic
negative. Visual Cognition, 12, 495–518.
Calder, A. J., Keane, J., Young, A., & Dean, M. (2000). Configural
information in facial expression perception. Journal
Experimental
Psychology:
Human
Perception
and
Performance, 26, 527–551.
Calder, A. J., & Young, A. W. (2005). Understanding the recognition of facial identity and facial expression. Nature
Reviews Neuroscience, 6(8), 641–651.
Capasso, R., & Miceli, G. (2001). Esame Neuropsicologico
per l’Afasia: ENPA (Vol. 4). Milano, Italy: Springer Science
& Business Media.
Capitani, E., & Laiacona, M. (1997). Composite neuropsychological batteries and demographic correction: Standardization
based on equivalent scores, with a review of published
data. The Italian Group for the neuropsychological study
of ageing. Journal of Clinical and Experimental
Neuropsychology, 19(6), 795–809.
Carlesimo, G. A., Caltagirone, C., Fadda, L., Marfia, G., Gainotti, G.,
& Gruppo per la standardizzazione della batteria per il

23

Deterioramento Mentale. (1995). Batteria per la valutazione
del Deterioramento Mentale (parte III): analisi dei profili qualitativi di compromissione cognitiva. Archivio di Psicologia,
Neurologia e Psichiatria, 4, 489–502.
Clarke, S., Lindemann, A., Maeder, P., Borruat, F. X., & Assal, G.
(1997). Face recognition and postero-inferior hemispheric
lesions. Neuropsychologia, 35(12), 1555–1563.
Crawford, J. R., Garthwaite, P. H., & Porter, S. (2010). Point and
interval estimates of effect sizes for the case-controls
design in neuropsychology: Rationale, methods, implementations, and proposed reporting standards. Cognitive
Neuropsychology, 27, 245–260.
Damasio, A. R., Damasio, H., & Van Hoesen, G. W. (1982).
Prosopagnosia Anatomic basis and behavioral mechanisms.
Neurology, 32(4), 331–331.
DeGutis, J., Cohan, S., Mercado, R. J., Wilmer, J., & Nakayama, K.
(2012). Holistic processing of the mouth but not the eyes in
developmental prosopagnosia. Cognitive Neuropsychology,
29, 419–446. doi:10.1080/02643294.2012.754745
DeGutis, J., Wilmer, J., Mercado, R. J., & Cohan, S. (2013). Using
regression to measure holistic face processing reveals a
strong link with face recognition ability. Cognition, 126(1),
87–100. doi:10.1016/j.cognition.2012.09.004
Della Sala, S., Laiacona, M., Trivelli, C., & Spinnler, H. (1995).
Poppelreuter-Ghent’s overlapping figures test: Its sensitivity
to age, and its clinical use. Archives of Clinical
Neuropsychology, 10(6), 511–534.
De Renzi, E., Faglioni, P., Grossi, D., & Nichelli, P. (1991).
Apperceptive and associative forms of prosopagnosia.
Cortex, 27(2), 213–221.
De Renzi, E., Motti, F., & Nichelli, P. (1980). Imitating gestures. A
quantitative approach to ideomotor apraxia. Archives of
Neurology, 37(1), 6–10.
Dodich, A., Cerami, C., Canessa, N., Crespi, C., Marcone, A.,
Arpone, M., … Cappa, S. F. (2014). Emotion recognition
from facial expressions: A normative study of the Ekman
60-faces test in the Italian population. Neurological Sciences,
35(7), 1015–1021.
Duchaine, B., & Nakayama, K. (2006). The Cambridge Face
Memory Test: Results for neurologically intact individuals
and an investigation of its validity using inverted face
stimuli and prosopagnosic participants. Neuropsychologia,
44(4), 576–585.
Duchaine, B. C., Parker, H., & Nakayama, K. (2003). Normal recognition of emotion in a prosopagnosic. Perception, 32, 827–
838. doi:10.1068/p5067
Duchaine, B., Yovel, G., & Nakayama, K. (2007). No global processing deficit in the Navon task in 14 developmental prosopagnosics. Social Cognitive and Affective Neuroscience, 2(2), 104–
113.
Eisenbarth, H., & Alpers, G. (2011). Happy mouth and sad eyes:
Scanning emotional facial expressions. Emotion (Washington,
D.C.), 11, 860–865. doi:10.1037/a0022758
Ekman, P., & Friesen, W. V. (1976). Pictures of facial affect. Palo
Alto, CA: Consulting Psychologists Press.
Farah, M. J. (1991). Cognitive neuropsychology: Patterns of cooccurrence among the associative agnosias: Implications for

24

C. MONTI ET AL.

visual object representation. Cognitive Neuropsychology, 8(1),
1–19.
Farnsworth, D. (1943). The Farnsworth-Munsell 100-hue and
dichotomous tests for color vision. JOSA, 33(10), 568–578.
Finzi, R. D., Susilo, T., Barton, J. J. S., & Duchaine, B. C. (2016). The
role of holistic face processing in acquired prosopagnosia:
Evidence from the composite face effect. Visual Cognition,
0(0), 1–17. doi:10.1080/13506285.2016.1261976
Fisher, K., Towler, J., & Eimer, M. (2016). Facial identity and facial
expression are initially integrated at visual perceptual stages
of face processing. Neuropsychologia, 80, 115–125. doi:10.
1016/j.neuropsychologia.2015.11.011
Fisher, K., Towler, J., & Eimer, M. (2017). Face identity matching
is selectively impaired in developmental prosopagnosia.
cortex, 89, 11–27.
Fortis, P., Maravita, A., Gallucci, M., Ronchi, R., Grassi, E., Senna, I.,
… Vallar, G. (2010). Rehabilitating patients with left spatial
neglect by prism exposure during a visuomotor activity.
Neuropsychology, 24(6), 681–697.
Fox, C.J., Hanif, H.M., Iaria, G., Duchaine, B.C., & Barton, J.J.S.
(2011). Perceptual and anatomic patterns of selective
deficits in facial identity and expression processing.
Neuropsychologia, 49, 3188–3200.
Gauthier, I., Behrmann, M., & Tarr, M. J. (1999). Is prosopagnosia
a general deficit in subordinate-level categorization. Journal
of Cognitive Neuroscience, 11, 349–370.
Gauthier, L., Dehaut, F., & Joanette, Y. (1989). The bells test: A
quantitative and qualitative test for visual neglect.
International Journal of Clinical Neuropsychology, 11(2), 49–
54.
Grüsser, O. J., & Landis, T. (1991). Faces lost: Prosopagnosia. Visual
Agnosias and Other Disturbances of Visual Perception and
Cognition: Vision and Visual Dysfunction, 12, 259–286.
Haxby, J. V., Hoffman, E. A., & Gobbini, M. I. (2000). The distributed human neural system for face perception. Trends in
Cognitive Sciences, 4, 223–233. doi:10.1016/S1364-6613
(00)01482-0
Hecaen, H., & Angelergues, R. (1962). Agnosia for faces (prosopagnosia). Archives of Neurology, 7(2), 92–100.
Hole, G. (1994). Configurational factors in the perception of
unfamiliar faces. Perception, 23(1), 65–74.
Humphreys, K., Avidan, G., & Behrmann, M. (2007). A detailed
investigation of facial expression processing in congenital
prosopagnosia as compared to acquired prosopagnosia.
Experimental Brain Research, 176(2), 356–373.
Kanwisher, N., McDermott, J., & Chun, M. (1997). The fusiform
face area: a module in human extrastriate cortex specialized
for the perception of faces. J. Neurosci, 17, 4302–4311.
Laiacona, M., Inzaghi, M. G., De Tanti, A., & Capitani, E. (2000).
Wisconsin card sorting test: A new global score, with Italian
norms, and its relationship with the Weigl sorting test.
Neurological Sciences, 21(5), 279–291.
Lee, Y., Duchaine, B., Wilson, H. R., & Nakayama, K. (2010). Three
cases of developmental prosopagnosia from one family:
Detailed neuropsychological and psychophysical investigation of face processing. cortex, 46(8), 949–964.

Maurer, D., Le Grand, R., & Mondloch, C. J. (2002). The many
faces of configural processing. Trends in Cognitive Sciences,
6(6), 255–260.
Mayer, E., & Rossion, B. (2013). Prosopagnosia. In O. Godefroy
(Ed.), The behavioral and cognitive neurology of stroke (pp.
231–232). Cambridge, UK: Cambridge University Press.
Mckone, E., & Yovel, G. (2009). Why does picture-plane inversion
sometimes dissociate perception of features and spacing in
faces, and sometimes not? Toward a new theory of holistic
processing. Psychonomic Bulletin and Review, 16(5), 778–
797. doi:10.3758/PBR.16.5.778
McNeil, J.E., & Warrington, E.K. (1993). Prosopagnosia: a facespecific disorder. Q. J. Exp. Psychol. A, 46, 1–10.
Meaux, E., & Vuilleumier, P. (2016). Facing mixed emotions:
Analytic and holistic perception of facial emotion
expressions engages separate brain networks. NeuroImage,
141(September), 154–173. doi:10.1016/j.neuroimage.2016.
07.004
Mestry, N., Donnelly, N., Menneer, T., & McCarthy, R. (2012).
Discriminating Thatcherised from typical faces in a case of
prosopagnosia. Neuropsychologia, 50(14), 3410–3418.
Minear, M., & Park, D. C. (2004). A lifespan database of adult
facial stimuli. Behavior Research Methods, Instruments, and
Computers, 36(4), 630–633. doi:10.3758/BF03206543
Monaco, M., Costa, A., Caltagirone, C., & Carlesimo, G. A. (2013).
Forward and backward span for verbal and visuo-spatial
data: Standardization and normative data from an Italian
adult population. Neurological Sciences, 34(5), 749–754.
Mondini, S., Mapelli, D., Vestri, A., & Bisiacchi, P. S. (2003). Esame
neuropsicologico breve. Milano: Raffaello Cortina Editore, 160.
Negrini, M., Brkic, D., Pizzamiglio, S., Premoli, I., & Rivolta, D.
(2017). Neurophysiological Correlates of Featural and
Spacing Processing for Face and Non-face Stimuli. Front.
Psychol, 8.
Palermo, R., Willis, M. L., Rivolta, D., McKone, E., Wilson, C. E., &
Calder, A. J. (2011). Impaired holistic coding of facial
expression and facial identity in congenital prosopagnosia.
Neuropsychologia,
49(5),
1226–1235.
doi:10.1016/j.
neuropsychologia.2011.02.021
Piepers, D. W., & Robbins, R. A. (2013). A review and clarification
of the terms “holistic,” “configural,” and “relational” in the
face perception literature. Frontiers in Psychology, 3(559),
1–11.
Ramon, M., Busigny, T., & Rossion, B. (2010). Impaired holistic
processing of unfamiliar individual faces in acquired prosopagnosia. Neuropsychologia, 48(4), 933–944.
Ramon, M., & Rossion, B. (2012). Hemisphere-dependent holistic
processing of familiar faces. Brain and Cognition, 78(1), 7–13.
doi:10.1016/j.bandc.2011.10.009
Rezlescu, C., Pitcher, D., & Duchaine, B. (2012). Acquired prosopagnosia with spared within-class object recognition but
impaired recognition of degraded basic-level objects.
Cognitive Neuropsychology, 29(4), 325–347.
Richler, J.J., Cheung, O.S., & Gauthier, I. (2011). Holistic processing predicts face recognition. Psychological science, 22(4),
464–471.

COGNITIVE NEUROPSYCHOLOGY

Riddoch, M. J., & Humphreys, G. W. (1993). Birmingham object
recognition battery. Hove, UK: Lawrence Erlbaum Associates.
Riddoch, M. J., Johnston, R. a., Bracewell, R. M., Boutsen, L., &
Humphreys, G. W. (2008). Are faces special? A case of pure
prosopagnosia. Cognitive Neuropsychology, 25(1), 3–26.
doi:10.1080/02643290801920113
Rivolta, D., Castellanos, N. P., Stawowsky, C., Helbling, S., Wibral,
M., Grützner, C., … Singer, W. (2014a). Source-reconstruction
of event-related fields reveals hyperfunction and hypofunction of cortical circuits in antipsychotic-naive, first-episode
schizophrenia patients during Mooney face processing.
Journal of Neuroscience, 34(17), 5909–5917.
Rivolta, D., Lawson, R. P., & Palermo, R. (2017). More than just a
problem with faces: Altered body perception in a group of
congenital
prosopagnosics.
Quarterly
Journal
of
Experimental Psychology, 70(2), 276–286.
Rivolta, D., Schmalzl, L., Coltheart, M., & Palermo, R. (2010).
Semantic information can facilitate covert face recognition
in congenital prosopagnosia. Journal of Clinical and
Experimental Neuropsychology, 32(9), 1002–1016. doi:10.
1080/13803391003662710
Rivolta, D., Woolgar, A., Palermo, R., Butko, M., Schmalzl, L., &
Williams, M. A. (2014b). Multi-voxel pattern analysis (MVPA)
reveals abnormal fMRI activity in both the “core” and
“extended” face network in congenital prosopagnosia.
Frontiers in Human Neuroscience, 8(November), 1–11.
doi:10.3389/fnhum.2014.00925.
Rondot, P., & Tzavaras, A. (1969). La prosopagnosie après vingt
années d’études cliniques et neuropsychologiques. Journal
de Psychologie Normale et Pathologique, 66, 133–165.
Rossion, B. (2008). Picture-plane inversion leads to qualitative
changes of face perception. Acta psychologica, 128(2), 274–289.
Rossion, B. (2013). The composite face illusion: A whole window
into our understanding of holistic face perception. Visual
Cognition, 21(2), 139–253. doi:10.1080/13506285.2013.772929
Scheller, E., Büchel, C., & Gamer, M. (2012). Diagnostic features
of emotional expressions are processed preferentially. PLoS
ONE, 7, e41792. doi:10.1371/journal.pone.0041792
Smith, M., Gosselin, F., Cottrell, G., & Schyns, P. (2005).
Transmitting and decoding facial expressions. Journal of
Vision, 4, 909. doi:10.1167/4.8.909.
Spinnler, H., & Tognoni, G. (1987). Standardizzazione e taratura
di test neuropsicologici. The Italian Journal of Neurological
Sciences, 8(Suppl. 6), 1–20.
Takahashi, N, Kawamura, M, Hirayama, K, Shiota, J, & Isono, O.
(1995). Prosopagnosia: a clinical and anatomical study of
four patients. Cortex, 31(13), 80365–80366.

25

Tanaka, J. W., & Farah, M. J. (1993). Parts and wholes in face recognition. The Quarterly Journal of Experimental Psychology, 46
(2), 225–245.
Tanaka, J. W., Kaiser, M. D., Butler, S., & Le Grand, R. (2012).
Mixed emotions: Holistic and analytic perception of facial
expressions. Cognition and Emotion, 26(6), 961–977. doi:10.
1080/02699931.2011.630933
Torfs, K., Vancleef, K., Lafosse, C., Wagemans, J., & de-Wit, L.
(2014). The Leuven perceptual organization screening test
(L-POST), an online test to assess mid-level visual perception.
Behavior Research Methods, 46, 472–487. doi:10.3758/s13428013-0382-6
Towler, J., Fisher, K., & Eimer, M. (2018). Holistic face perception
is impaired in developmental prosopagnosia. Cortex, 108,
112–126.
Tranel, D., Damasio, A. R., & Damasio, H. (1988). Intact recognition of facial expression, gender, and age in patients with
impaired recognition of face identity. Neurology, 38(5),
690–690.
Van Belle, G., de Graef, P., Verfaillie, K., Busigny, T., & Rossion, B.
(2010). Whole not hole: Expert face recognition requires holistic perception. Neuropsychologia, 48, 2620–2629.
Warrington, E. K., & James, M. (1991). The visual object and space
perception battery. Bury St Edmunds: Thames Valley Test
Company.
White, D., Rivolta, D., Burton, A. M., Al-Janabi, S., & Palermo, R.
(2017). Face matching impairment in developmental prosopagnosia. The Quarterly Journal of Experimental Psychology,
70(2), 287–297.
Yin, R. K. (1969). Looking at upside-down faces. Journal of
Experimental Psychology, 81(1), 141–145.
Young, A. W., Hellawell, D., & Hay, D. C. (1987). Configurational
information in face perception. Perception, 16, 747–759.
doi:10.1068/p160747
Young, A. W., Newcombe, F., de Haan, E. H., Small, M., & Hay, D.
C. (1993). Face perception after brain injury. Selective impairments affecting identity and expression. Brain, 116, 941–959.
doi:10.1093/brain/116.4.941
Young, A. W., Perrett, D., Calder, A., Sprengelmeyer, R., &
Ekman, P. (2002). Facial expressions of emotion: Stimuli and
tests (FEEST). Bury St. Edmunds: Thames Valley Test
Company.
Zoccolotti, P., Antonucci, G., Judica, A., Montenero, P.,
Pizzamiglio, L., & Razzano, C. (1989). Incidence and evolution
of the hemi-neglect disorder in chronic patients with unilateral right brain-damage. International Journal of
Neuroscience, 47, 209–216.