A review of the impairments, preserved visual functions, and neuropathology in 21 patients with visual form agnosia – A unique defect with line drawings

We present a comprehensive review of the rare syndrome visual form agnosia (VFA). We begin by documenting its history, including the origins of the term, and the first case study labelled as VFA. The defining characteristics of the syndrome, as others have previously defined it, are then described. The impairments, preserved aspects of visual perception, and areas of brain damage in 21 patients who meet these defining characteristics are described in detail, including which tests were used to verify the presence or absence of key symptoms. From this, we note important similarities along with notable areas of divergence between patients. Damage to the occipital lobe (20/ 21), an inability to recognise line drawings (19/21), preserved colour vision (14/21), and visual field defects (16/21) were areas of consistency across most cases. We found it useful to distinguish between shape and form as distinct constructs when examining perceptual abilities in VFA patients. Our observations suggest that these patients often exhibit difficulties in processing simplified versions of form. Deficits in processing orientation and size were uncommon. Motion perception and visual imagery were not widely tested for despite being typically cited as defining features of the syndrome – although in the sample described, motion perception was never found to be a deficit. Moreover, problems with vision (e.g., poor visual acuity and the presence of hemianopias/ scotomas in the visual fields) are more common than we would have thought and may also contribute to perceptual impairments in patients with VFA. We conclude that VFA is a perceptual disorder where the visual system has a reduced ability to synthesise lines together for the purposes of making sense of what images represent holistically.


Introduction
Sensory agnosia refers to a selective, modality-specific disorder of perception that cannot be attributed to a primary sensory deficit, inattention, or general mental impairment (for a review on a general history of sensory agnosia, see Coslett, 2011). The aim of the present review was to provide a synthesis of all reported cases of "visual form agnosia" (VFA). The term was first used by Benson and Greenberg (1969) to describe a patient who had a "unique defect in visual shape perception" (pg. 82). The label has since been applied to several other patients who presented with similar symptoms. However, there is some disagreement about the exact nature of these symptoms, including which patients have the syndrome. This review documents the reported features of the syndrome and discusses the case studies in the literature who meet those criteria. We begin with a brief history of VFA. This is followed by a discussion of how VFA has been conceptualised in other articles. In this discussion, we propose that it is useful to draw a distinction between 'shape' and 'form' 1 in understanding the nature of VFA as a perceptual disorderwhich is something that has not really been considered before. Individual case studies are then described, including how symptoms were tested for. We conclude with a discussion on broad similarities, notable areas of divergence, and considerations for future research.

Brief history
VFA is conceptualised as a type of apperceptive agnosia. The origins of this term can be traced to the end of the 1800s, when some researchers were actively trying to identify and describe visual impairments. Hughlings Jackson's work on imperception (1876), Munk's work (1881) on seelenblindheit ('mindblindness') in dogs, and Freud's (1891) introduction of the term agnosia were all important contributions to the field. However, it was the work of Lissauer in 1890, dubbed the father of agnosia, which stood the test of time (Riddoch and Humphreys, 2003). Lissauer wrote about a patient, Gottlieb L., who presented with a strange perceptual complaint. Although this person could still clearly see, in that they could copy figures presented to them, and describe their general features and configurations, they were completely unable to identify the same figures. For example, Lissauer writes the following regarding Gottlieb's reaction when shown a pocket watch: "A lamp", after a bit "a lighter". On request draws the watch clearly. The watch is then put to his ear. He recognises it at once: "Oh, it's a watch" (pg. 175, Lissauer and Jackson, 2007;Lissauer, 1890).
Thus, despite being able to see and draw the pocket watch, only after hearing the ticking could he correctly identify it. Lissauer labelled this impairment associative agnosiathe person was unable to associate their prior knowledge with intact visual perception. Other types of visual agnosia specific to other sensory modalities (e.g., audition) were considered possible. Lissauer speculated on theoretical grounds that there could be another form of visual agnosia, one where the structural encoding of the sensory stimulus is disrupted with a resulting failure to perceive objects. This he termed apperceptive agnosia, which remained a theoretical possibility until 1918, when Goldstein and Gelb published their case study on patient Johann Schneider.
At 24 years old, Schneider suffered a head injury after a minesplinter exploded. Schneider displayed several neuropsychological impairments following this damage and became a controversial case in neurology. Schneider could identify images and words slowly but only when allowed to trace them with his head and hand. Otherwise, they were unidentifiable. His visual fields were constricted but visual acuity and other visual functions were adequate, including colour vision with only a slight red/green anomaly. This pattern of symptoms led Goldstein and Gelb to conclude that Schneider's deficits were an example of apperceptive agnosia and a striking example of defective Gestalt perception, i.e., the inability to group single elements of a composite visual image and segregate figure from ground in static visual displays (Heider, 2000). Due to the apparent strangeness of Schneider's symptoms, he appeared in more than a dozen works in the years after, presenting with "alexia, form agnosia, loss of movement vision, loss of visual imagery, tactile agnosia, loss of body schema, loss of position sense, acalculia, and loss of abstract reasoning" (pg., 633, Marotta and Behrmann, 2003). Questions were then raised about the legitimacy of Schneider's symptoms, as one person displaying all these symptoms at once is remarkable (for a detailed discussion, see Landis et al., 1982). For example, Jung (1949) saw the patient and noted that the stereotypical head tracing movements were present but were overstressed during replication of the Goldstein and Gelb tests and disappeared with increasing interest in the material. Thus, Jung suggested that some of the abnormal responses that made Schneider's case study interesting may have been overlearned or exaggerated. Nevertheless, Schneider was the first example of apperceptive agnosia, and thereafter, more case studies were described wherein there were clear visual impairments not attributable to a low-level sensory deficit.
It soon became clear that there were distinct types of apperceptive agnosia, and that Lissauer's two-category scheme appeared to be too narrow to capture the nuanced differences between patients and the different kinds of apperceptive agnosia. The highly specific agnosia for shape or form did not appear, or was not labelled, until 1969 by Benson and Greenberg, when they described Mr S. As described below, Mr S appeared to have profound problems with the recognition of basic shapes and line drawings, despite sufficiently near normal visual acuity and visual fields. The label was then retroactively applied to one other case, HC (Adler, 1944), and there have been several more cases reported since. Note that although Benson and Greenberg were the first to use the label in its current form, Nielsen (1936) appears to have predated this when he discussed the most fundamental of visual agnosias in a comprehensive review. Nielsen termed it geometric-optic agnosia. "In this difficulty, the patient has lost sense of direction of lines so that he fails to recognise objects because of their distortion … In geometric-optic agnosia other mental disturbances are usually present." (pg. 52). There are many individuals who appear to meet this criterion, as we will discuss later.
It deserves passing mention that some people have questioned if form impairments, like those exhibited by patients with VFA, can be classified as part of the agnosia continuum or should be considered a low-level sensory problem (Bay, 1953;Warrington and James, 1988). Indeed, this led some authors, such as Warrington and James (1988), to argue that cases like Mr S cannot be considered examples of apperceptive agnosia as described initially by Lissauer for two key reasons: (1) it is difficult in cases of visual agnosia to unequivocally establish that visual functions (e.g., visual acuity, visual fields etc.) are normal or sufficiently near normal, and (2) the term agnosia ought to be reserved for problems with knowledgeafter all, the Greek term from which agnosia is derived translates roughly to 'lack of knowledge'. In other words, instead of a conceptual problem, or the imposition of meaning onto sensory data, as in Lissauer's case of associative agnosia, the symptoms of VFA may represent more of a pseudo-agnosia, laying somewhere between low-level sensory problems and higher-order aspects of agnosia. This scepticism has contributed to confusion as to how one can operationally define apperceptive agnosia and by extension VFA. In the present paper, we adopt Farah's (2004) more general definition of apperceptive agnosia and consider VFA as an example of it. Namely, apperceptive agnosia is a broad category referring to "any failure of object recognition in which perceptual impairments seem clearly at fault, despite relatively preserved sensory functions such as acuity, brightness discrimination, and colour vision" (pg. 11). VFA is a specific instance of this where shape or form perception is disrupted.

What are the defining features of the syndrome?
VFA is primarily characterised by deficits in the perception of shape or form. Indeed, coinage of the term 'form agnosia' by Benson and Greenberg was intended to capture deficits in both shape and form, given they never made a distinction between the two. In the visual arts, the former is not the same as the latter (Art in Context, 2022). After careful consideration, we believe making this distinction is helpful for discussing patient symptoms and evaluating how the brain processes visual information. Unfortunately, as evidenced from conducting this review, the word 'form', as in everyday parlance, has been used in a variety of ways when discussing VFA and rarely operationally defined, which seems to have contributed to some confusion and a lack of consensus as to what the syndrome could be.
The visual arts are careful to draw a distinction between shape and form (Art in Context, 2022;Arnheim, 1974). For a stimulus to have form, it must have a global configuration (e.g., an outline or boundary of some kind) and an arrangement of local features within that boundary to give an impression of a three-dimensional structure. The most common type of stimulus shown to patients when they are being examined to assess form perception are line drawings. Form in a 2D drawing simulates structure as it is encountered in the real world. This structure is conveyed by the addition of pictorial depth cues, such as occlusion, shading, and texture. Hence, a 2D image can be thought of as a shape when it comprises only outlines, whereas a 2D image has form when it gives the illusory impression of having a third dimension. For example, a square in a picture has shapehaving height and width. Conversely, a cube has formhaving length, width, and height (see Fig. 1). In everyday parlance, there is little distinction between shape and form, with the two often being used interchangeably. We believe it is useful to differentiate between the two to help understand the nature of deficits in VFA.
From a phenomenological perspective, reports from VFA patients range from "everything blurs together" (DF, in Milner et al., 1991) to "I can't tell what an object is because I don't see it clearly" (Mr S, in Efron, 1969). This qualitative account of perception can help explain why there are profound deficits in shape or form perception (Heider, 2000). VFA patients experience seeing objects that lack definable form and are unable to perceive their figural properties or their 'traits and lines' (Martinaud, 2017). This has led some to posit that they are also unable to perceive the metric features of objects, such as their size (e.g., lengths and distances) and orientation (Riddoch and Humphreys, 2003;Martinaud, 2017). These deficits are usually demonstrated by patients being unable to discriminate between basic shapes, such as a square from a circle, and copying and identifying drawings successfully.
Typical tasks used to assess difficulties in shape or form perception include the Efron task (Efron, 1969) and the Boston Naming Test (Kaplan et al., 2001). The Efron shape matching task was originally designed to determine if Mr S was able to discriminate between two objects that differed only by shape whilst matching for volume. In this task, the patient is presented with two objects at oncea black square and a black rectangle with identical volume and reflectance. Within the pair, the shape of one of the objects is progressively elongated (see Fig. 2). The task requires the participant to determine if the two are the same or different. The Boston Naming Test (Kaplan et al., 2001) consists of 60 black-and-white line drawings of objects that must be named. Successful naming is assumed to necessitate shape and form processing. Although these are some of the typical tests used, it is more common for researchers to develop their own tasks for evaluating form perception. Doing so often yields interesting experimental findings but makes it difficult to directly compare task performance across studies. Finally, the recognition of real objects is generally better than line drawings (Farah, 2004). This is thought to be due to patients focusing on what they can still see, such as colour and texture, in images.
VFA patients retain several visual functions. However, it is misleading to say that they are usually normal. Rather, it may be more prudent to say that these functions might not be sufficiently affected to explain the agnosia. Visual fields are largely intact, or, if there are any visual field defects (e.g., scotomas), they do not seem responsible for the perceptual problems (Farah, 2004). When measured, visual acuity is often reported to be within the normal range. As discussed in the proceeding sections, however, this is not entirely accurate in many cases.
Often, the reported metrics of visual acuity deviate considerably from what most people would consider normal (i.e., 20/20), even though the authors purport otherwise.
Measuring visual acuity is made difficult by the fact that some patients cannot recognise letters, making the Snellen chart an inappropriate test for measuring visual acuity. Instead, the testing of visual acuity sometimes involves assessing the patient's ability to differentiate small dot patterns. For example, in patient DF, her visual acuity seemed intact as she could distinguish between a grey patch and a fine dot pattern (26 dots per cm) that is equivalent to a resolution of 1.7 min arc . So, despite subjectively stating that things might blur together, a wide range of ophthalmological tests in patient DF suggest there is no defect that would be sufficient to account for her complaints of unclear vision.
In most VFA patients, the ability to maintain fixation, colour perception, stereopsis, motion, and brightness perception are considered sufficiently intact (Farah, 2004). Thus, it is through compensatory processing strategies, like processing colour and texture in objects, that patients can display better recognition for real objects in pictures compared to line drawings (Mapelli and Behrmann, 1997). For example, a patient's explanation when asked how they correctly identified a lion was: "I can see it's yellow and furry, so I just guessed." This is not to say that their perception of figures is normal, as they often make decisions with great difficulty and take much longer than the average observer. So, while they can still see some things and have normal or near-normal vision, without the capacity to perceive form or shape configuration, profound problems with visual recognition occur. Finally, although not To the left, we have a cube with shading cues. To the right, we have the shading cues removed from the cube, which leaves the impression of a flat shape.

Fig. 2.
These illustrations depict the Efron task, which consist of presenting pairs of rectangles that differ in dimensions but not in surface area. Sizes are adapted from the LEA Rectangles Game (https://www.leatest.com/catalog/ cognitive-vision/lea-rectangles-game%C2%AE). VFA patients are often unable to perform better than chance when discriminating whether the pairs are same or different. a visual function per se, Farah (2004) observed that some VFA patients can utilise kinaesthetic cues to facilitate object recognition, such as tracing a figure with their fingers or head.
There appears to be a variety of causes for VFA. Heider's review (2000) included only patients who suffered from carbon monoxide poisoning, where all individuals had extended lesions in the extrastriate cortex, particularly the lateral occipital complex (LOC). However, this only comprises a subset of patients with VFA. Farah (2004) notes that the neuropathology in cases of VFA is fairly homogenous. She describes five patients who suffered from carbon monoxide poisoning, one from mercury poisoning, and another from a penetrating head wound. 2 Neurological signs and neuroimaging techniques suggest that brain damage in most patients is primarily located in the occipital lobe and surrounding regions. However, brain damage tends to be diffuse rather than focalparticularly in cases with carbon monoxide poisoning, which is known to also damage subcortical white matter and the cortex diffusely (Farah, 2004). Even Farah's detailed investigation overlooked some existing cases in the literature, and some new cases have since been published, which further complicates the question of what brain regions might be implicated. As such, a revisiting of the literature is necessary. Below, in Tables 1 and 2, we detail all cases post 1940 which either have been classified as VFA patients or, who upon reading their reported symptoms, display one or several of the defining features for VFA. These cases were identified by an initial search in PubMed with the search terms "visual form agnosia", which returned 12 unique cases from 77 results, and "apperceptive agnosia", which returned 9 unique cases from 33 results. The remaining cases were identified through reading review articles and case study papers that refer to other cases of VFA. In total, we identified 31 cases with symptoms of VFA, 10 of which were excluded for various reasons, including comorbid diagnoses or insufficient details on the presence of symptoms to confirm the possible presence of VFA. More information on excluded cases can be found in the Appendix.

Review of patients
Existing works on VFA and the criteria detailed in them paints a relatively coherent picture of the syndrome. However, upon reading case study papers directly, reported deficits and patterns of brain damage have sometimes been described in unhelpful ways for the purposes of drawing similarities across studies. For example, possible nuanced comparisons between patients are lost when authors provide gross as opposed to detailed descriptions. Nuances are important because they can lead one to better characterise the nature of the syndrome. It is our opinion that there needs to be a single place where all cases are described in enough detail to evaluate their similarities and disparities, as well as the tests that were administered to evaluate their symptoms. Therefore, this review aims to document how well the 21 case studies that we found meet the defining criteria for VFA. They are presented below in the order in which they were published. In addressing this aim, we comprehensively describe deficits, intact capabilities, and neuropathology on a case-by-case basis, including how they were examined. It should be noted that many original papers do not mention the specific tests or measures used. We have included this information whenever possible. We place emphasis on addressing the following questions: Where, if mentioned, is the site of brain damage? How was the brain damage acquired? Do patients have intact visual fields and visual acuity? Are patients able to discriminate between objects based on shape or form or both? Are patients able to perceive orientation, size, colour, and motion? Can patients copy figures? Is imagery preserved? Can patients still read? Tablea 1 and 2 lists each VFA patient that we review along with their demographics, descriptions of brain damage, and presence and absence of visual abilities. Figs. 3 and 4 provide illustrated tallies of this same information. Adler (1944) and Sparr et al. (1991) HC sustained carbon monoxide poisoning in a nightclub fire at the age of 22. She was first described by Adler (1944) and is considered the first genuine case of VFA to appear in the literature after Schneider. As such, she has been foundational to the study of VFA. HC was tested again nearly 50 years later by Sparr et al. (1991). She provides unique insight into how symptoms can progress and change over an extended period.

Impaired abilities
In the first month after brain damage, HC was unable to recognise simple shapes ("What is this?" [a circle] "Looks like an alphabetical A."). She did not recognise any objects from pictures (e.g., she called a comb a fountain pen and a little toy elephant a pencil). Although severely affected to begin with, she gradually developed the ability to copy simple shapes (e.g., circles, triangles, and squares). She had greater difficulty copying figures with curved lines. Pictures were harder to recognise than real objects. Three months after brain damage, HC became more adept at recognising one part of an image to infer what the whole image represented. For example, when shown a green 4-inch-long toy battleship through a tachistoscope, she gave the following responses for different viewing presentation times: 1 s -"a fountain pen"; 2 s -"a knife, green"; 3 s -"a boat". She explained her method as the following: "First I saw the front part. It looked like a fountain pen because it was shaped like a fountain pen. Then it looked like a knife because it was so sharp … Then I saw the spokes and thought it was shaped like a boat." Forty years later, HC had no difficulty identifying common objects. It was only upon formal testing that her agnosia became evident. HC could only truly perceive a small portion of any image. From those smaller portions, she inferred what the image represented using a 'sum-of-parts' strategy. This strategy was confirmed by her inability to recognise figures in the Poppelreuter test, shown in Fig. 5. Synthesising small stimuli, complex details, and curves seemed more affected than straight lines. Regarding imagery, in 1944 the visual component of her dreams was diminished, but she had not lost the ability to visualise or form new images as evidenced by her drawing ability. In 1991, imagery was reported to be defective, as free drawing remained crude, she was unable to describe objects adequately, and topographical imagery was reported as impaired.

Intact abilities
In 1944, visual acuity was 20/70 in each eye, as inferred by the Snellen chart. Despite a slight constriction of the right lower visual field, her visual fields were deemed normal. It was reported that her colour perception was intact, though slightly perturbed. HC could provide good estimates of object length, reflecting intact size perception, and read arrows, like those posted at her hospital, reflecting intact orientation perception. More rigorous tests of orientation and size perception were not performed. In 1991, HC's corrected visual acuity was 20/30 in each eye, as inferred by the Snellen chart. Goldmann perimetry was performed and was deemed normal aside from a small homonymous scotoma subtending the midline of the lower visual field, which encroached the central 10 • of vision inferiorly. It was reported that HC performed normally on colour naming and matching tasks. Sparr et al. (1991) performed electroencephalography (EEG) and magnetic resonance imaging (MRI) on HC, which revealed bilateral occipital atrophy. No other pathology was reported. Age denotes the age of the patient when they were first examined. Abbreviations other than patient initials: BA = Brodmann area, CO = carbon monoxide, CT = computer tomography, EEG = encephalography, F = female, LOC = lateral occipital complex, M = male, MRI = magnetic resonance imaging, and PET = position emission tomography. V1, V2, V3, V4, and V5 are different visual areas. Benson and Greenberg (1969) and Efron (1969) At 25 years of age, Mr S was found unconscious after being exposed to leaking carbon monoxide fumes while showering. This was the case study where the term visual form agnosia was coined and is perhaps the most severe case of VFA in this review.  Fig. 3. The figures represent tallies of cases with different types of brain damage and visual impairments. Panel A depicts the number of cases with damage to a particular lobe in the brain. Panel B depicts the laterality of damage either being bilateral, right hemisphere (RH) only, left hemisphere (LH) only, or inconclusive (IC). Panels C to G depict the presence and absences of various visual abilities. For the latter panels, green denotes the number of cases with preserved abilities, red denotes the number of cases with impaired abilities, and yellow denotes where the information provided was too inconclusive for us to decide whether these abilities were preserved or impaired.

Impaired abilities
Mr S could not identify objects by sight alonethis included physical objects, pictures of objects, body parts, letters, numbers, and geometrical figures. For example, he consistently failed to identify, copy, and match simple figures and letters. He also maintained that he experienced no dreams, or if there were dreams, they were devoid of visual content, suggesting abnormal visual imagery.

Intact abilities
Mr S could successfully name colours. He could use colour and size information to try to identify stimuli. For example, when he was shown a set of keys, he said that he could see "it" while pointing in its general direction but was only able to report it as "shiny like silver". When the experimenter jingled the keys, however, he snapped out "keys" (Efron, 1969). Visual acuity was estimated to be 20/100 in each eye, as assessed by his ability to reach for objects or point to small pieces of white cardboard on a black background. Performance on other physiological tests of vision, like spatial summation and flicker fusion, was reported to deviate from normal only slightly (Efron, 1969). Some bilateral inferior visual field constriction was reported.

Brain damage
An EEG revealed a persistent bilateral slow wave pattern in the parietooccipital areas, while a pneumoencephalogram revealed posterior ventricular dilation bilaterallysuggesting damage to both the Fig. 4. The figures represent tallies of cases with different types of visual impairments. Namely, panels A to G depict the presence and absences of various visual abilities. For the latter panels, green denotes the number of cases with preserved abilities, red denotes the number of cases with impaired abilities, and yellow denotes where the information provided was too inconclusive for us to decide whether these abilities were preserved or impaired. striate cortex and non-primary visual cortex in the occipital and parietal lobes. Benson and Greenberg posited that "the isolated loss of form discrimination suggests malfunction at a higher (cortical) level, a premise consistent with the recognised pathology of carbon monoxide intoxication" (1969, pg. 87). They noted that the presence of other problems, like gait disturbance and the inability to recognise orally spelled words, indicated that brain damage was likely more extensive than damage confined to visual regions in the occipital and parietal lobes. (2000) and Campion and Latto (1985) RC suffered carbon monoxide poisoning from being trapped inside a chimney. He was examined on several occasions between 1979 and 1985. Campion and Latto (1985) posited the "peppery mask hypothesis". They argued that VFA in RC was the result of 'peppery' scotomas throughout his visual field, resulting in disproportionate effects on form perception over other kinds of perception. Note this explanation can only apply to RC and other VFA patients with 'peppery' scotomas. It cannot account for VFA in patients with intact visual fields or other forms of scotomas.

Impaired abilities
RC was unable to recognise and copy simple shapes or trace their outline with his finger. He was unable to name these items, describe their function, or make same/different judgements between them when presented with pairs. He could also not recognise the orientation of lines and grating patterns. When asked to name 30 line-drawings of objects from the Boston Naming Test and 27 real world equivalents of the drawings, he scored 5/30 for the line drawings and 17/27 for the real objects. Discriminating line length was impossible. RC performed poorly when asked to detect differences in luminance and desaturated coloursdemonstrating that colour and brightness perception was also impaired.

Intact abilities
It was reported that RC's visual fields were intact aside from a right lower homonymous quadrantanopia. Sensory visual functions were reported to be intact, although more specific details were not reported.

Brain damage
Computed tomography (CT) revealed a diffuse lesion located in and above the left calcarine sulcus, encompassing the striate and extrastriate cortex. There was also some damage in the extrastriate cortex in the right hemisphere. Landis et al. (1982) Mr X was exposed to inorganic mercury vapour in a laboratory job. Landis et al. (1982) note the high correspondence between the symptom profile of Mr X and patient Schneider, including their use of kinaesthetic tracing to facilitate recognition.

Impaired abilities
When allowed to trace, Mr X could recognise simple shapes, such as circles and triangles. However, with more complex figures, he would give different answers for the same drawing depending on where he started to trace. Recognition of detailed pictures was easier than line drawings. He could not draw, copy, and recognise forms, even when he was allowed to trace them and was given unlimited time. Visual fields were reported to be restricted but normal within 2 • radius, as assessed using tachistocopic perimetry testing. Mr X was said to have deficits with visual imageryalthough what tests were used to determine this was not reported.

Intact abilities
Colour vision was reported to be normal as assessed by the 100 Hue Farnsworth Test and Ishihara Isochromatic Plates. Stereoscopic vision was also considered normal using random dot stereograms. Ample evidence of being able to detect motion was also provided. It was reported that Mr X's visual acuity was 20/20 in each eye.

Brain damage
The following was reported: "computerized tomography of the head, electroencephalography and electromyography/nerve conduction studies were all within normal limits" (Landis et al., 1982). Although no brain damage appeared from these examinations, Landis et al. (1982) speculated that Mr X could have had a considerable degree of posterior white matter and callosal destruction. The authors reasoned that inorganic mercury vapour is known to poison cellular function in the brain (Bernhoft, 2012) and discussed the results from a brain autopsy of a different patient who suffered from mercury poisoning (Hay et al., 1963). In this patient, there was a porous appearance of white matter in several brain regions, most notably subcortical white matter tracts in the temporal and occipital lobes, as well as the corpus callosum. According to Landis et al. (1982), Mr X could have had similar brain damage. Alexander and Albert (1983) ES was found unconscious from carbon monoxide inhalation in a burning home and remained unconscious for 2 days after the incident. Although ES was first described by Alexander and Albert (1983), he was not considered a VFA patient until Farah (2004) suggested that he might be. There is little detail about which tests were used to assess his perceptual abilities.

Impaired abilities
Performance was reduced on many tasks, including reading letters and numbers, identifying faces, and recognising geometric figures and complex figures. He could not draw or copy any of the line drawings presented to him.

Intact abilities
Visual acuity could not be measured by conventional means due to ES' problems with letter recognition. Nonetheless, he could see small objects, suggesting that visual acuity was somewhat intact. ES' visual fields and extraocular movements were reported to be normalalthough he had difficulties maintaining fixation. He could no longer see objects when fixation was lost. He could name colours and real objects presented physically in front of him.

Brain damage
EEG was reported to be normal. A CT scan revealed moderate enlargement of cortical sulci. From this scan, Alexander and Albert (1983) concluded that ES had widespread cortical damage, including damage throughout the occipital lobes. Milner et al. (1991)

and others
At 34-years-old, DF suffered from carbon monoxide toxicity due to a leaking gas water heater while showering. She is the most studied of all VFA patients, appearing in over 58 publications by Melvyn Goodale, David Milner, and their colleagues. Her brain damage has been documented thoroughly by James et al. (2003) and Bridge et al. (2013).

Impaired abilities
DF has reduced abilities in object recognition. As discussed by Milner et al. (1991), she cannot recognise pictures of some objects but can identify others based on their colour, texture, and size (e.g., she can identify a screwdriver by first recognising it as being 'long, black, thin, and metallic'). She cannot identify Snodgrass and Vanderwart (1980) line drawings at all. DF also performs poorly on the Efron task. Namely, she cannot perceptually discriminate between two different rectangular stimuli with the same surface area. DF was described as partly impaired with motion processing.

Intact abilities
Although she cannot perceptually discriminate between two stimuli in the Efron task, she can calibrate her peak grip aperture in flight during reach-to-grasp actions with high-levels of precision to match the width of Efron blocks (Goodale et al., 1991). In other words, processing the form of these objects for the purposes of perception is impaired but not for the purposes of grasping them. Importantly, this behaviour cannot be explained by compensatory mechanisms given that the calibration of her peak grip aperture always occurs in flight before she touches the objects and can still occur with the rest of her body remaining immobile. This led Goodale and Milner (1992) to propose their two visual streams hypothesis (TVSH) where vision for processing the form of stimuli is carried out independently by the ventral and dorsal visual streams for the purposes of perception and action, respectively. This dissociation has been replicated multiple times with other stimuli (e.g., posting a letter in a slot that can be oriented in different orientations, Milner et al., 1991; grasping Blake shapes with stable versus unstable grasp points, Goodale et al., 1994) in controlled settings with appropriate control conditions to rule out alternative explanations than TVSH. DF's visual fields are intact except for some constrictions in her upper right quadrant. Other aspects of vision, such as luminance, brightness judgements, acuity, stereoscopic depth, and colour perception, are normal as assessed by computerized psychophysics and ophthalmological examination. In terms of visual imagery, DF could draw reasonably well from memory, which is sometimes taken as evidence of intact imagery capabilities. James et al. (2003) showed there was a concentration of bilateral damage in the lateral occipital complex (LOC) in the ventral stream, with sparing of the fusiform gyrus and primary visual cortex in the two hemispheres. A later study by Bridge et al. (2013), built on the earlier work of James et al. (2003), provides a more in-depth structural account of her brain. Bridge et al. (2013) showed that there was a substantial loss in cortical thickness in LOC as well as reduced white matter connections between LOC and other areas. Areas that were minimally damaged in DF include V1, MT/V5, and the ventromedial cortex in the temporal lobes. Both James et al. (2003) and Bridge et al. (2013) noted some damage in the left parietal-occipital cortex. Davidoff and Warrington (1993) SMK was assaulted at 21 years of age and suffered anoxia for an extended period. SMK is not considered by some to be a case of VFA, given that he retains the ability to recognise a range of visual stimuli.

Impaired abilities
On informal tests, Davidoff and Warrington (1993) reported that SMK "totally misidentified many objects in his immediate environment and he failed to recognise familiar faces. He was able to identify only one (chair) of the first six of the Oldfield pictures. He had considerable difficulty in identifying simple clear silhouette drawings of common objects (3/12 correct) (pg. 84)." Reading abilities could not be assessed. SMK had specific problems with shape discrimination, performing poorly on a modified version of the Efron task examining his ability to discriminate between solid objects and collinear segments. Size discrimination was also impaired, where he performed at chance when asked to point to the larger of two stimuli when they differed by only 10% in area.

Intact abilities
SMK's visual acuity was within the normal range on the Ffooks symbols test (1965). His subtle problems with shape perception did not permeate all aspects of visual recognition: surface characteristics were accurately recognised (e.g., colour, brightness, and shading), figures occluded with visual noise were detected, and he could perceive Gestalt laws of proximity, continuity, and closure. Despite displaying some difficulties with shape processing, he performed perfectly when asked to point to the square or triangle in a forced choice task that involved judging the position of a Kanizsa illusory shape presented randomly on either the left or right visual fields. SMK could also recognise geometric forms (e.g., cubes) but line drawings of objects (e.g., Snodgrass stimuli) were not tested.

Brain damage
The authors conjectured that SMK had a lesion in and around the human homologue of V4, as they reasoned that a similar pattern of impaired discrimination abilities had been reported in monkeys after large lesions to V4 (Heywood and Cowey, 1987). There is no neuroimaging data available to confirm this speculation. Shelton et al. (1994) FWT had a stroke at 66 years old. He was initially blind. His vision gradually improved, as assessed by examination, even though he stated that he could not see because everything seemed to "run together".

Impaired abilities
FWT had bilateral superior quadrantanopia for moving and stationary targets but could consistently identify the on and off-set of a light flash in all visual fields. The patient failed to name, point to, or match items of a specific colour. Simple shapes could not be named (0/9 correct), pointed to (0/9 correct), or copied (0/9 items initially correct and 3/9 items correct at a 9-month follow up) but could be matched to a sample (10/12 correct). Snodgrass and Vanderwart (1980) line-drawings could not be named (0/30 correct) or copied (0/30 correct) but the majority could be matched (28/40 correct).

Intact abilities
Snellen charts could not be used to assess visual acuity due to letter recognition failure. However, he was able to indicate that all 14-point numbers on the Snellen chart were different, suggesting visual acuity at 20/100 in each eye. Initially, he was unable to discriminate between line orientations that differed less than 90 • . At a follow-up evaluation, he scored 75% correct on a task requiring him to make same-different judgements for lines differing in orientations by 18 • -90 • . FWT consistently discriminated between black dots that differed by 2-mm in diameter. FWT could trace simple forms except for their upper portion, which is in line with his visual field defects. He could trace the missing parts of the stimuli when they were inverted vertically. FWT was reasonably good at describing objects visual features from memory, suggesting intact visual imagery. Interestingly, FWT was reported to also use a tracing technique, like Mr X, to facilitate recognition (Shelton et al., 1990).

Brain damage
A CT scan showed a lesion in the inferior temporal lobes bilaterally. The lesion was more severe in the left hemisphere and encompassed the fusiform and lingual gyri. The primary visual cortex was spared. A positron emission tomography (PET) scan was also performed. This scan corroborated the findings of the CT scan. Okuda et al. (1996) KK was admitted to hospital at 56-years-old because of a transverse myelitis after receiving a clinical diagnosis of multiple sclerosis at 23. KK is the only organic case study with VFA where symptoms did not develop because of trauma.

Impaired abilities
KK's accuracy ranged from 60 to 80% when asked to identify 15 real objects on three separate occasions. Her perception of line drawings was more severely impaired than real objects, with KK correctly identifying 38% of eight line drawings. Recognition was poor in overlapping Poppelreuter figures (1/4 correct) and photographs of objects (0/12 correct). KK's ability to name colours was also poor, averaging scores of 2/8 and 5/8 correct on the first and second testing sessions, respectively.

Intact abilities
KK had a right-sided homonymous hemianopia. Visual acuity was reported to be 20/60 in the right eye and 20/100 in the left eye when corrected with reading glasses. KK could copy simple figures such as a cross, circle, and triangle. Compared to her deficits with object recognition, KK performed better at reading letters and words in her native Japanese language. She could correctly read 80% of Kanji letters and 65% of Kana letters. Like FWT, KK could describe objects reasonably well from memory, suggesting intact visual imagery.

Brain damage
MRI showed white matter damage in the frontal and occipital lobes in the two hemispheres, as well as in the corpus callosum. Her lesions in the occipital lobe extended anteriorly to the occipitotemporal junction. KK also had a frontal-lobe biopsy, which revealed demyelination and gliosis. Charnallet et al. (1996) PG sustained head trauma at 27 years of age and was in a coma for three weeks. PG has perceptual problems that are specific to elementary shape recognition.

Impaired abilities
PG could not copy lines presented at different orientations. He could not trace or describe them either. He was unable to compare the length of lines or the size of circles, as assessed by the Birmingham Object Recognition Battery (BORB) subtests 2-3 (Riddoch and Humphreys, 2022), identify fragmented letters, or discriminate between basic shapes on the Efron task. PG also performed poorly with recognition of Snodgrass and Vanderwart (1980) line drawings (196/258 correct) and silhouettes (0/20 correct).

Intact abilities
Although PG performed poorly on tasks that required him to perceive lines, shapes, and line drawings, he performed much better on recognition tasks involving real objects (30/30 correct) and detailed images (photographs: 32/32 correct; realistic drawings: 127/152 correct). Visual testing revealed a partial left homonymous hemianopia. It was reported that PG had perfect central vision, normal visual acuity, normal ocular movement control, and showed no signs of ocular ataxia.

Brain damage
An MRI showed bilateral frontal lobe lesions, diffuse lesions of the white matter, and a large cortical-subcortical lesion in the occipital and parietal lobes in the right hemisphere. Vecera and Gilds (1997) and others JW suffered a cardiac event while exercising, which led to a state of anoxic encephalopathy. For whatever reason, JW is not widely cited as a case of VFA, despite there being clear evidence of deficits in shape processing.

Impaired abilities
JW recognised 21 of 39 real objects presented to him visually. He could recognise 38 of 39 of them when they were presented to him by touch or sound. His recognition of black-and-white drawings was worse, naming 2 of 34 randomly selected Snodgrass and Vanderwart (1980) drawings. He was unable to perform simple visual image segmentation, the Efron task, and shape detection tasks consisting of objects presented against a background of visual noise. He reported being a 'poor reader'although no formal testing of reading was mentioned.

Intact abilities
Tests of his vision were normal aside from the presence of an upper left quadrantanopsia detected with Goldmann perimetry. His visual acuity was poor, estimated to be 20/200 in each eye (Mapelli and Behrmann, 1997). His colour vision was intact as inferred by the Farnsworth-Munsell 100-hue test. Mapelli and Behrmann (1997) noted that JW could reliably differentiate between basic shapes with gross differences (e.g., a square from a circle) but struggled to differentiate between shapes that differed more finely (e.g., a circle vs. an oval). His struggles on the Efron task described earlier could relate to difficulties differentiating minor differences between shapes. Unpublished results from Mapelli and Behrmann are cited stating that imagery was intact in JW.

Brain damage
CT scans obtained shortly after his incident revealed V1 and V2 injury. Rosenthal and Behrmann (2006) further reported that he had bilateral lesions in the ventral occipital lobes, areas that provide inputs to LOC. Grossman et al. (1997) At 54, a massive myocardial infarction resulted in SZ being admitted to hospital. Like JW, SZ does not appear in any previous review papers on VFA.

Impaired abilities
SZ could not read any words at all (0/15 correct). When asked to name black-and-white drawings from the Boston Naming Test, SZ accurately named 5 of 60 pictures. On a match-to-sample paradigm, SZ was unable to judge if two partially overlapping shapes were the same or different, being accurate 31% of the time. When asked to match simple designs composed of shapes (e.g., a square with a line extending perpendicularly from the middle of one side) according to orientation, SZ could match 43% of items differing in orientation by 90 • or less, and 50% of items rotated 180 • . Objects in colour photographs were never correctly identified and attempts at identification typically focused on colour cues. For example, when shown a picture of a room, SZ responded by saying: "I can see orange, I can see black: it must be a Halloween picture.'' The colours were correct, but the picture was not a Halloween scene" (pg. 323, Grossman et al., 1997). Real objects were recognised somewhat betteralthough still poorly. When shown 20 real objects, SZ was accurate on only 4 trials.

Intact abilities
Simple shapes (e.g., circles, squares, rectangles) could be discriminated with 80% accuracy in a match-to-sample paradigm. Although this is not normal performance, it is still well above chance. The ability to name objects by touch or sound was 100% accurate. Colour perception was intact. SZ was 100% accurate judging colours in a match-to-sample task. Identifying colour blobs was used to assess SZ's visual acuity, which was estimated to be 20/40 (presumably in each eye, although this is not stated). SZ's visual imagery is reported as being better than their visual perception, as he could draw well from memory, and verbal description of object features were considered good.

Brain damage
An MRI scan 1 month after his accident was normal. However, a 6month follow-up MRI scan revealed an abnormally high signal intensity in the occipital association regions in the two hemispheres. A PET scan at this same time "revealed extensive hypoperfusion bilaterally in middle and inferior temporooccipital cortices that spared primary visual cortex" (Grossman, et al., 1996). Ferreira et al. (1998) The patient was a 65-year-old retired engineer who had a left occipitotemporal haemorrhage in 1981 and a right occipitotemporal haemorrhage 1991.

Impaired abilities
The patient exhibited achromatopsia, as assessed by the Ishihara test (specific scores were not provided). Additionally, it was reported that visual field testing revealed a bilateral quadrantanopia in the right inferior and left superior quadrants. The patient's ability to name Snodgrass and Vanderwart (1980) line drawings was impaired (3/122 correct) and their performance in determining whether line drawings corresponded to real objects or non-objects was also deficient (31/40 correct) and did not improve with the use of silhouettes of the same stimuli (26/40 correct). Furthermore, the patient's ability to recognise real objects was impaired when presented statically (7/30 correct) or in motion (3/30 correct).

Intact abilities
It was reported that visual acuity was normal in the preserved visual fields. Pupillary responses were also considered normal. The patient could produce accurate copies of Snodgrass and Vanderwart line drawings and performed normally on Benton's line orientation test (23/ 30 correct). The patient could identify real objects by touch and could describe and pantomime their use, confirming that they knew what they were.

Brain damage
MRI images are provided in the case report demonstrating damage to the occipitotemporal cortex in the two hemispheres. No other damage is reported.
2.14. SF in Aglioti et al. (1999) At 40 years old, SF suffered cardio-respiratory arrest during surgery. Consequently, he developed bilateral posterior brain atrophy. This is another case that has been overlooked in the literature. This may be because, like ES, there are few details about the tests employed to verify impairments.

Impaired abilities
It was reported that SF could not identify or discriminate simple visual shapes or objects. SF was acutely aware of this impairment. It was reported that he consistently failed at standard clinical tests of reading and visual form recognition, and that he could not recognise single black letters on a white background.

Intact abilities
SF could name colour stimuli without difficulty. He could recognise the direction of motion and judge the distance of visual targets based on binocular and monocular vision. Visual imagery was largely normal. There is no mention of visual acuity or visual fields being assessed.

Brain damage
MRI performed six months after his trauma showed posterior brain atrophy that was most marked in the occipital and parietal lobes. Lê et al. (2002) SB developed visual agnosia at 3 years old following meningoencephalitis. He was examined by Lê et al. (2002) much later in life, at the age of 30.

Impaired abilities
Orientation perception was reported to be disrupted. This impairment was inferred from his performance on two tasks. The first assessed his abilities to detect an oddly oriented line amongst a matrix of many lines and the second consisted of a BORB subtest that assessed line orientation perception. His performance on the latter comprised 18/30 items correct. It was also reported that SB was achromatopsic. He could not discriminate, order, or name colours, and he failed the Farnsworth-Munsell 16-Hue test. Standard drawings of common objects could not be identified, as assessed by the BORB. Although SB made several errors recognising real everyday objects, he was able to make intelligent guesses based on the features of the objects presented to him. Real objects were recognised better than photographs, achieving accuracy scores of 35/45 and 8/45, respectively. SB's shape perception tests showed abnormal results, but the abnormality was due to his slower speed of processing rather than any issues with accuracy. He performed at chance levels when discriminating between overlapping figures. Goldmann perimetry revealed a left lateral homonymous hemianopia, which spared the macular. The authors noted that this examination was difficult to administer because SB had trouble maintaining central fixation. Common clinical tests of visual acuity could not be performed due to SB being unable to read letters. Instead, the authors administered a same/different test where letters or shapes were presented at different retinal sizes. From this test, acuity was estimated to be 4/10 with both eyes without correction.

Intact abilities
Stereopsis, brightness judgements, luminance detection, and motion processing were all reported to be normal. Despite being unable to recognise drawings, SB was able to copy them wellfor example, he could copy the Rey-Osterrieth figure. SB performed well on BORB subtests assessing his abilities to match line lengths (23/30 correct) and stimulus sizes (27/30 correct). Like DF, SB was able to use vision for action and performed similarly to DF on tests of visuomotor control. For example, the slot test , and grasping Blake shapes (Goodale et al., 1994). Finally, in terms of imagery, SB was normal on several tests including mentally comparing the size of different objects, and mentally comparing a series of items and finding a similarity between them based on some visual property (e.g., global shape). Drawing from memory, as well as describing objects from memory was also relatively good. Finally, motion processing was deemed largely normal with an extensive battery (opto-kinetic nystagmus, moving dots displays, moving sine wave gratings). SB reported no difficulty in everyday motion processing like estimating the speed of walking persons or moving cars.

Brain damage
MRI revealed lesions in the occipitoparietal and occipitotemporal regions in the right hemisphere, and in the occipitotemporal junction in the left hemisphere. Right hemisphere lesions included partial or complete damage to V2, V3, V4 and V5 (MT), with a sparing of V1. In the left hemisphere, lesions comprised the ventral part of the occipitotemporal junction. In short, the left and right ventral streams, as well as the right dorsal stream, were damaged. Hildebrandt et al. (2004) AM had heart arrest at 46 years old. He was in a coma for three days after the incident. AM does not appear in any other discussion of VFA other than the Hildebrandt et al. (2004) paperdespite the authors describing the patient as having 'visual form agnosia'.

Impaired abilities
The authors administered the Visual Object and Space Perception (VOSP) battery and the BORB. In doing so, they found impaired performance on the identification of line drawings (14/40 correct). AM was below the cut-off for normal performance on tasks assessing his abilities in length matching (18/30 correct), size matching (22/30 correct), silhouette recognition (recognising only 4 items correctly, cut-off for normal performance being 16), and object recognition (recognising only 2 items correctly, cut-off for normal performance being 15). Some aspects of motion processing were impaired, as inferred by moving dot displays.

Intact abilities
The Goldmann perimetry test did not reveal any visual field defects. AM could identify illusory contours, Kanizsa figures, and colour. Using piecemeal strategies, AM could copy the Rey-Osterrieth figure. He made only a few errors on the Gailinger copy test, which required him to connect dots to draw a shape. AM could perform the orientation matching task in the BORB. In addition, he could read single letters provided they were sufficiently spatially apart and read words provided they were in fonts that lacked serifs (e.g., in 'Arial' as opposed to 'Times Roman') and were sufficiently large (i.e., >16 pt).

Brain damage
There was no structural damage evident on an MRI taken two months after his heart attack. Barton et al. (2004) Patient 006 was a 52-year-old man when he was examined 7 months after a right medial occipito-temporal haemorrhage. His haemorrhage occurred during a surgical resection of an oligodendroglioma. The paper's primary focus was on patient 006's prosopagnosia. Later, Karnath et al. (2009) referred to the patient as also having visual form agnosia based on his difficulties perceiving non-face stimuli, as reported by Barton et al. (2004).

Impaired abilities
In addition to patient 006's difficulties recognising faces, patient 006 performed poorly on the Benton orientation task that required him to match line segments varying in spatial orientationachieving a score of 13/30 correct. Patient 006 also had difficulties with silhouettes on the VOSP object perception tasks. Issues with curvature perception were also reported.

Intact abilities
Snellen acuity was 20/25 in each eye. Goldmann perimetry revealed a complete left homonymous hemianopia. He correctly identified 12 of 14 Ishihara pseudoisochromatic plates. Eye movements were reported to be normal. On the Ghent overlapping figures test, patient 006 achieved a score of 55/57 correct, demonstrating an ability to recognise line drawings. Patient 006 could also read, achieving a score of 49/50 correct on the Warrington word task (Warrington, 1984).

Brain damage
MRI showed damage to the right medial occipital cortex comprising the lingual and fusiform gyri. Yang et al. (2006) XF was 42 years old when she was examined by Yang et al. (2006). She suffered from carbon monoxide poisoning due to a fire and regained consciousness 20 days after the incident. XF has not appeared in any discussions of VFA other than in Yang et al. (2006).

Impaired abilities
XF had difficulty reporting the correct orientation of sinusoidal gratings presented in either horizontal, vertical, or diagonal positions. She reported all of them as being in the vertical orientation. XF was also impaired in discriminating object size. In addition, she was unable to perform Gestalt grouping tasks based on similarity, proximity, continuity, or symmetry. XF performed poorly when tasked to name shapes (scoring 30% correct) or indicate whether two shapes were the same or different (scoring 56% correct). She could not copy these same shapes. When presented with Snodgrass and Vanderwart line drawings, XF correctly named 11% of coloured objects and 2% of black-and-white line drawings.

Intact abilities
Static perimetry revealed normal visual fields. Acuity could not be assessed through letter reading but was reported to be normal as assessed using contrast sensitivity functions. Brightness perception was reported to be relatively preserved. XF was 95% accurate on a colour naming task. In terms of motion perception, XF was able to judge the direction of a moving target correctly ~90% of the time.

Brain damage
MRI revealed enlarged ventricles and damage to the occipitotemporal and anterior frontal cortex in the two hemispheres. Occipitotemporal lesions in the two hemispheres included LO and the posterior part of the fusiform gyrus. Her early visual cortex, including the primary visual cortex, was reported to be relatively intact. Her MRI also revealed damage in the left posterior parietal cortex. Other parts of the parietal cortex were spared. Riddoch et al. (2008) SA suffered a stroke at 50-years old. The trauma left her with difficulties in object recognition and reading. She appears in Riddoch et al. (2008) where her symptom profile was compared to one with a patient with integrative agnosia. 3 SA has not appeared in any subsequent review articles on VFA despite the authors reporting that she had VFA.

Impaired abilities
Object recognition was assessed using Snodgrass and Vanderwart (1980) images. She correctly named 155/260 (60%) images. Additionally, she named 10/15 animate and 3/15 inanimate images of coloured objects. Shorter exposure times exacerbated her recognition abilities. There was some evidence of decreased contrast sensitivity, especially for higher contrasts. On a modified Efron task, she scored 10/10 correct for the easiest discrimination level, 9/10 correct for the next level of difficulty, and 5/10 correct for the hardest level of difficulty. SA performed worse naming overlapping compared to non-overlapping figures.

Intact abilities
Judgements of line length, line orientation, and circle size on the BORB exam were all within the normal range based on established norms. Visual acuity with correction was normal, as assessed with the typical Snellen chart. SA was able to read quite well, scoring 46/50 on the Warrington Recognition Test (Warrington, 1984). SA appeared to be able to copy drawings, albeit poorly (see Fig. 3a in Riddoch et al., 2008;pg., 64). In addition, she could discriminate figure-from-ground. Like AM, she could reproduce a decent copy of the Rey-Osterrieth Figure (see Fig. 3b in Riddoch et al., 2008;pg., 64). 3 Defined as "a case of apperceptive agnosia who succeeded on shape discrimination and shape-copying tasks, but who failed on more stringent tests of visual perception (such as distinguishing the individual items in an overlapping figures test, or detecting targets embedded in displays of homogenous distractors)" (Humphreys et al., 2008).

Brain damage
MRI revealed a lesion encompassing the dorsal extrastriate cortex, including the intraparietal sulcus, in the right hemisphere. The lateral occipital region and the striate cortex were spared. Karnath et al. (2009) JS suffered an ischemic stroke during a coronary angiography. After the procedure, he complained that he could not see, despite being able to safely navigate in his environment. For instance, he could not recognise objects, watch TV, or read newspapers, but he could take regular walks, frequenting his local grocery store for shopping. Many of the experiments performed by Karnath et al. (2009) were adopted from previous experiments performed by Melvyn Goodale, David Milner, and colleagues in DF. This was done to evaluate JS's abilities to process form for the purposes of action as well as for the purposes of perception.

Impaired abilities
JS was severely impaired at the recognition of objects. He recognised only 3 out of 12 real objects, even without time constraints. He also failed to recognise any of the fragmented and completed line drawings from the Fragmented Picture Test (Kessler et al., 1993). Again, this poor performance was measured without time limits. Similar results were found on the Boston Naming Test. He could identify only 2 of 15 line drawings. Copying shapes (e.g., circle, square) but not simple line drawings (e.g., comb, racket) was good. Perceptual matching of two slots (like DF in Milner et al., 1991) in an orientation judgement task was poor relative to controls. Perceptual discrimination of Blake shapes was again inferior to healthy controls too (like DF in Goodale et al., 1994)although his responses were 73% correct, which was above chance levels.

Intact abilities
Tests used to assess basic visual function were not specified. However, the authors reported that there were no visual field defects. Colour perception was reported to be largely normal based on JS only making 2 errors on a colour discrimination task. Like DF, JS showed normal vision for action on the slot task and grasping irregular Blake shapes.

Brain damage
MRI revealed a bilateral lesion to the fusiform and lingual gyri, extending into the adjacent posterior cingulate gyrus. In the right hemisphere, the lesion also extended into the parahippocampal gyrus and cuneus. Lateral cortical structures of the occipital and temporal lobes were intact, which differs from DF with extensive damage to LOC. Serino et al. (2014) SDV suffered an electrocution-induced heart-attack. As a result, he acquired brain damage that left him completely blind for almost 10 months. His vision progressively improved but he never regained his abilities to recognise common objects, faces, and words. Serino et al. (2014) evaluated SDV 3 years after his heart attack.

Impaired abilities
Copying forms, such as a cube, was impaired. On BORB subtests, SDV performed poorly (53% correct) on the size matching task. He was unable to perform the line-matching and figure recognition tests either. He was impaired recognising overlapping shapes. SDV was unable to read strings of letters, including real words and non-words. SDV had a central visual field deficit with relative sparing of the periphery. Specifically, he was blind in most of the fovea (i.e., within 5 • of eccentricity) but could progressively see closer to 10 • of eccentricity. Namely, SDV could roughly discriminate between vertical versus horizontal line orientations at different locations in his peripheryalbeit his performance was worse compared to healthy controls.

Intact abilities
SDV could copy simple shapes (5/7 correct) and draw basic shapes from memory (3/3 correct). SDV accurately named 10 colours presented as solid rectangles. SDV's performance assessing hue discrimination between two colours did not differ from healthy controls. His performance on modified random dot kinematograms also did not differ from healthy controls. Visual imagery was reported to be normal as evidenced by his abilities to recall perceptual details of common stimuli correctly. For example, he could indicate correctly whether a particular animal had a long or short tail (17/17 correct) and whether they had upward or downward ears (13/13 correct). He could also indicate correctly whether a letter had curved or straight lines (11/11 correct). Visual acuity could not be assessed by normal measures, so a contrast sensitivity task was used instead to assess his spatial acuity. SDV's contrast sensitivity profiles were identical to those obtained in control participants after correcting for deficits with orientation perceptionindicating that he had good visual acuity.

Brain damage
MRI revealed enlarged ventricles and two lesions in the occipital lobe bilaterally. Specifically, there was evidence of brain damage in Brodmann areas (BA) 17 and 18, and partial damage in BA 19. SDV's brain damage also extended rostrally to the superior parietal lobes, corresponding to brain damage in BA 30-31 and partial brain damage in BA 7.

Discussion
The present review aimed to document the impairments, intact abilities, and neuropathology of all patients with VFA that we could find. This was done to collate all cases in one place and provide a comprehensive overview of how symptoms were tested. We now provide a discussion on similarities and disparities between patients. We then note broader methodological lessons learned from this exercise. A retrospective analysis on this kind of historical data is challenging because the data are relatively sparse and there is no way to include additional clarifying information. Nevertheless, having the information presented in a single location affords the researcher their own reading and comprehension of the source material. The key takeaways from our review are: (1) damage often occurs in multiple areas, with damage more often occurring in the ventral visual stream; and (2) impairments are relatively specific to line drawings of objects, and this likely reflects a processing problem associated with the synthesis of lines in complex stimuli, causing a break down in perception. We build upon key findings in the following sections of the discussion.

Location of brain damage
All patients had damage to the occipital lobes, aside from AM. This is not surprising given the visual nature of the symptoms. Interestingly, the primary visual cortex seemed preserved in most patients. Only Mr S, RC, JW, and SDV were reported to have damage to this regionalthough some additional cases were described as having generalised atrophy of the occipital lobe, which may have included damage to the primary visual cortex. Nonetheless, strong arguments could be made that LOC is critical for form processing and that bilateral damage to this region leads to VFA. There exists an abundance of neuroimaging evidence demonstrating that LOC is strongly and usually activated when neurologically intact participants visually process shape, line drawings, and other forms (Grill-Spector et al., 1999;Kourtzi and Kanwisher, 2001;Denys et al., 2004;Sayim and Cavanagh, 2011;Chouinard et al., 2008;Peel and Chouinard, 2022). In addition, DF has bilateral damage to LOC and has profound deficits perceiving the shape and form of objects. However, conclusions of damage to LOC as the sole cause of VFA cannot be made for multiple reasons.
First, conclusions that can be made about the precise areas of damage in the patients we reviewed are limited. Although it is noteworthy that most patients had bilateral lesions (17/21), many did not have detailed structural or functional neuroimaging data available. Therefore, it is unclear how many of them had damage to LOC. Second, LOC appears to have been spared in some cases. For example, JS had a profound VFA, yet it was reported that his LOC was intact in both hemispheresalthough there was bilateral damage to his adjacent fusiform gyrus (Karnath et al., 2009). Finally, the aetiology leading to brain damage and VFA varied considerably across patients, further complicating matters for localising a particular brain region that could underpin the disorder. These aetiologies comprise carbon monoxide poisoning (6), mercury poisoning (1), anoxia (5), stroke (5), organic development with multiple sclerosis (1), head trauma (1), meningoencephalitis (1), and oligodendroglioma (1) -typically varying in the degree of focalised brain damage. For example, carbon monoxide toxicity typically results in diffuse patterns of brain damage (Farah, 2004). Therefore, it is unlikely that damage to LOC is the sole cause of VFA. What is perhaps more likely is that VFA arises from a disruption in the flow of information through the ventral visual stream from the primary visual cortex to LOC and other higher-order areas implicated in form processing (Tanaka, 1996). A similar idea was put forth by Karnath et al. (2009) based on their evaluation of JS. Alternatively, LOC is the most critical area in the brain for form processing and any damage to its inputs or outputs or other areas that LOC depends on for it to process form will result in VFA.

Visual acuity and visual fields
Many case reports asserted that basic vision was either normal or that any problems with basic vision, such as reductions in visual acuity or visual-field defects, could not be the cause of perceptual deficits in their VFA patients. In terms of the latter, Bender and Feldman (1972), and more recently Serino et al. (2014), have argued that accompanying alterations in basic vision are not always inconsequential to shape and form perception. We share their concerns. In this review, we noted that there were significant problems in visual acuity in 3 patients (Mr S, JW, and FWT) and inconclusive reports of visual acuity in 7 patients (JS, ES, KK, SF, AM, XF, and SB). Hence, only half of the VFA patients we reviewed can be considered as having normal visual acuity.
In terms of visual fields, normal visual field perimetry was only found in 5 patients (Mr X, ES, SMK, AM, and JS) (24%). The other 16 patients (76%) were reported to have defects in their visual fields, such as a scotoma or a hemianopia. The presence of normal visual field perimetry in the 5 patients with VFA is not explained by a common factor. These patients had different aetiologies, including mercury poisoning, carbon monoxide poisoning, anoxia, and trauma. Therefore, it is challenging to pinpoint a specific mechanism of action that can explain why these 5 patients and not the others had intact basic vision. Similarly, among the 16 patients with reported abnormalities, 8 displayed unilateral deficits, while the remaining 8 had bilateral deficits. Again, the variety of damage patterns and aetiologies make it difficult to establish a precise mechanism of action as to why these 16 patients had problems with basic vision.
Hence, the widely held notion that VFA patients have intact basic vision is simply not true. Most have problems. This is an important consideration. In a case study of a patient with a quadrantanopia, it was observed that their scotoma was affecting how visual stimuli were perceived in their intact and inside their missing visual field (Dilks et al., 2007). Namely, the patient perceived shapes as being elongated near and inside their scotoma. Specifically, the patient reported seeing a rectangle and an ellipse when a square and a circle were actually presented to them, respectively. Consider how a problem like this (i.e., the perceived elongation of a visual stimulus near and inside a scotoma) might impact a patient's ability to perform the Efron task, which requires making same or different judgements about a progressively elongated shape relative to another. Subtle distortions in size would have profound consequences when differentiating pairs of objects in the Efron task, such as those shown in Fig. 2. It could be the case that some of the patients in this review could have had low-level visual problems mimicking VFA and may not actually have had VFA. For this reason, patients like DF, which have been examined extensively and whose basic visual functions remained largely intact, provides critical validation to VFA.

Shape and form perception
The most common symptom appears to be poor recognition of line drawings of objects. SMK was not assessed in recognising line drawings of objects and only patient 006 demonstrated intact abilities in this domain. All other patients demonstrated impairments either copying, matching, or naming line drawings. Stimuli used to evaluate these abilities varied considerably, comprising Snodgrass and Vanderwart (1980) images, images from the Boston Naming Test, images from the Poppelreuter and Ghent overlapping figures, images from VOSP subtests, images from the Fragmented Picture Test, and stimuli created by the researchers. A potential explanation for patient 006's preserved abilities could be that he was the only patient whose damage was confined to one hemisphere. It was reported that patient 006 had damage in the right but not the left hemisphere. In this situation, visual areas in the intact hemisphere could stand in for homologous areas that are damaged in the affected hemisphere. Nonetheless, patient 006 did have difficulties identifying silhouettes, demonstrating problems in global shape processing.
Line drawings can sometimes require both shape and form processing given that form can be conveyed by occlusion depth cues (e.g., the wings of an airplane shown in front or behind its fuselage when the stimulus is viewed sideways). An interesting lesson from our review is that we cannot confidently say whether shape processing was impaired in many patients when this processing was properly examined in isolation by the Efron task, the VOSP subtests, or the use of line drawings confined to simple shapes. This is surprising when one considers that the term 'form agnosia' was originally intended to describe shape processing deficits by Benson and Greenberg (1969). We can only confidently say that 10 of the 21 cases had impaired shape processing abilitiesnamely, Mr S, RC, ES, DF, PG, SF, XF, SA, JS, and SB.
We deemed the remaining 11 cases as inconclusive. HC displayed impairments in shape processing for only the first month after brain damage, but the patient showed normal performance in subsequent testing sessions. Mr X could recognise shapes when given an unlimited exposure time and allowed to trace them, but his performance dropped when tracing was disallowed. SMK performed poorly on the Efron task but could recognise Kanizsa shapes, overlapping shapes, and simple geometric forms. FWT could copy basic shapes but not more complicated ones. KK could copy shapes but could not identify them. JW could reliably differentiate between line drawings of shapes (e.g., a square from a circle) but performed poorly on the Efron Task. SZ was 80% accurate at identifying line drawings of simple shapes but fell to 31% accuracy when an additional line was introduced into the images. "Not named" was not tested on basic shape processing. AM struggled to recognise silhouettes but could copy the Rey-Osterrieth figure and recognise Kanizsa figures, overlapping, and simple geometric forms. Patient 006 struggled to identify silhouettes but could successfully recognise Ghent overlapping figures. SDV and KK could copy simple shapes but struggled copying more complicated ones.
Thus, simple shape processing impairments do not appear to be ubiquitous across VFA patients. A deficit with more complex line drawings is more common. Why might this be the case? As raised in the Introduction, it might help to consider shape and form as conceptually different in a similar way as artists do. From an artistic perspective, a 2D image of a shape comprises an outline devoid of apparent depth whereas a 2D image of a form comprises a shape with additional cues giving it an apparent 3D structure. A square is an example of a shape. It is flat and restricted to two dimensions (i.e., height and width). Conversely, a cube is an example of a form. It has three dimensions (i.e., length, width, and height) (see Fig. 1). Indeed, equating shape and form as the same has led some to refer to VFA as "impaired shape processing (form agnosia)" (Gerlach and Robotham, 2021; Benson and Greenberg, 1969) -which highlights how previous research in this space has not considered them different. Differentiating shape and form, as we are doing here, is important given this historical context. Finding that only half or so patients displayed deficits with processing basic shapes suggests that statements to the effect of 'participants cannot differentiate a square from a circle' may be inaccurate or even overstated. Our main contention therefore is that there is a qualitative difference conveyed between shapes and forms in 2D images, where shapes are basic, and forms are more complex.
Functional neuroimaging provides evidence that shape and form, when conceptualised as different constructs, could be processed by different neural substrates. Although Kourtzi and Kanwisher (2001) demonstrate that the LOC processes both shape and form, other studies examining other areas in the brain provide evidence for different populations of neurons in nearby regions (e.g., V3A) that fulfil a greater role processing shapes than forms (Denys et al., 2004;Grill-Spector et al., 1998). This difference in neural substrates could explain why shape processing is disrupted in some but not all patients. More widespread brain damage might result in deficits with both shape and form processing while less widespread brain damage may result in deficits that affect shape or form processing more than the other.
Thinking of form from an artistic perspective can also clarify why processing line drawings seems consistently affected in VFA. Depictions of form simulates three-dimensional structure more than depictions of shape. In the case of the former, the impression of three-dimensional structure is achieved through processing global and local properties of stimuli that are defined by edges and lines. In the real world, there are no lines around objects. Instead, the structure of objects is conveyed through discontinuities in brightness, textures, colours, and other visual qualities. In drawings, lines are configured to convey structure in 2D spacesome simulating three-dimensional structure, or form, more effectively than others. This pictorial form can range from what we call simple to complexas illustrated in Fig. 6. The star (a) does not have any pictorial form. It is a shape. The bananas (b) comprise mostly of outlines, but they also have other lines and occlusion cues to provide apparent depth to simulate more three-dimensional structure. The bananas can be said to have simple pictorial form. The early workings of the portrait (c) give a richer impression of three-dimensional structure than the bananas (b) with the addition of even more types of depth cues. The portrait can be said to have complex pictorial form. Line drawings from the Boston Naming Test and the Snodgrass and Vanderwart (1980) set lie somewhere in between these simple and complex pictorial forms, and patients with VFA perform poorly on recognition tasks involving these stimuli that have a limited impression of three-dimensional structure but are at a higher level of complexity compared to basic shapes. Therefore, we think it is useful to conceptually differentiate shape and form, although it is not something that is typically done.
How does this perspective fare with contemporary theories of shape processing? We suggest that contemporary theories of shape are largely synonymous with our definition of form. Namely, "global shape representations are achieved by describing objects via the spatial arrangement of their local features rather than by the appearance of the features themselves" (Ayzenberg and Behrmann, 2022). From this perspective, shape encompasses local features and their spatial arrangement which are key to conveying an object's structure. Such descriptions typically do not draw a distinction between shape and form in 2D images, but we feel the importance of processing local cues in addition to global ones is demonstrated as being deficient in the VFA patients we reviewed.
Indeed, one quibble that we have with this perspective which largely equates shape and form (e.g., Ayzenberg and Behrmann, 2022) is that it somewhat overlooks the importance of local feature appearances to structure. This is best exemplified in one aspect of shape within this framework, which is of the shape skeleton (Blum and Nagel, 1978;. The shape skeleton models object structure by describing the spatial arrangement of contours and component parts via internal symmetry axes. It has been demonstrated that this description can help determine an object's structure even with noisy or incomplete contour information (Ayzenberg et al., 2019). While this approach seems reasonable for basic 2D shapes, we suggest that when dealing with more complex forms, processing the appearance of internal features within the boundary is important and can lead to qualitatively different perceptions (see Fig. 1). The limitations of the shape skeleton metaphor, especially when applied to complex images, have been discussed by Arnheim (1974) with the famous duck-rabbit image. This image presents a drawing that can be perceived as either the head of a duck facing left or a rabbit facing right. Although the image possesses only one quantitative structure and, consequently, one shape skeleton, the processing of internal features within the boundary can result in two equally valid structural interpretations. This example emphasises the significance of processing internal features alongside global structure and highlights the potential for qualitatively different perceptual outcomes.
All this discussion concerns flat two-dimensional images. A reasonable question then follows: How does this problem with form relate to abilities in recognising objects in the real world? Consider the following studies that examined the recognition of both line drawings and real objects in VFA patients. Mapelli and Behrmann (1997) aimed to address why JW's identification deficits were more pronounced with black-and-white (B/W) line drawings compared to real objects. To this end, they created two versions of line drawings of 39 common objects: one in B/W and the other in colour. Abilities to recognise these drawings were compared to the real versions of the same objects. JW recognised the real versions (21/39 correct) better than either the B/W (7/39 correct) or coloured (12/39 correct) line drawings, which demonstrates that he could recognise stimuli better when they provide more information about structure. Drawings that are richer in apparent structure achieved from the addition of shading and textures seem to be recognised better in VFA patients. Indeed, PG was better at recognising more realistic drawings (i.e., like the one in panel c of Fig. 6) compared to more basic ones from the Snodgrass and Vanderwart (1980) set (i.e., like the one in panel b of Fig. 6) (84% and 76%, respectively).
However, these results are not always apparent in VFA patients. Holler et al. (2019) found that images of complex pictorial forms are not recognised efficiently in DF and JW. Holler and colleagues examined object recognition in the following display formats: real objects in both DF and JW, high-resolution images of the same objects in both DF and JW, and stereoscopic images of the same objects with a virtual third dimension in JW only. For the latter, active shutter glasses were used to present two images of an object from two slightly different viewpoints to the left and right eyes to create stereoscopic depth. Both patients performed poorly recognising the high-resolution images and JW performed poorly recognising the stereoscopic images. Recognition performance improved modestly in both patients with the real objects. The authors of the study referred to this latter effect as a 'real object advantage' (Holler et al., 2019). Nonetheless, one should keep in mind that the real object advantage was modest. The patients' performance with the real objects remained worse than control participants, as evidenced by higher error rates and longer response times. Interestingly, JW performed worse in recognising both the high-resolution images and the 3D stereoscopic images compared to the real objects, which suggests that artificially creating stereoscopic depth does not completely simulate three-dimensional structure as in the real world.
In the patients that we cover in our review, we found real object advantage in 9 patients (HC, RC, ES, DF, KK, PG, JW, SZ, SB) (43% of patients), a real object deficit in 6 patients (Mr S, Mr X, NN, SF, AM, JS) (29% of patients), and inconclusive real object advantage in 6 patients (SMK, FWT, 006, XF, SA, SDV) (29% of patients). Real object advantage is a promising avenue for further research that could potentially by informative. The ecological validity of line drawings is questionable given they vary in the degree to which they simulate structure. One can even argue that images of any kind cannot fully simulate structure. Pictorial cues that artists apply to produce apparent depth, such as shading, occlusion, and texture gradients, conflict with other cues telling the brain that the eyes are looking at a flat surface, such as stereopsis, vergence, motion parallax, and accommodation. Virtual threedimensional space created by stereoscopic images reduces but does not eliminate this conflict. Conflicting depth information from a mismatch in vergence and accommodation is an issue in virtual reality (Hoffman et al., 2008). Other differences in processing real objects versus images have been demonstrated by Snow and colleagues (e.g., Snow et al., 2011;Snow and Culham, 2021) and our research group (e. g., Kithu et al., 2021). However, this issue not new. Notably, James Gibson (1979) seriously questioned the ecological validity of presenting stimuli as images in his book The Ecological Approach to Visual Perception.
In short, we suggest that deficits in processing and synthesising local and global features for the purposes of perceiving structure might contribute to symptoms of VFA. The proposed definition of VFA distinguishes itself from conditions like simultanagnosia and integrative agnosia in terms of the underlying difficulties in processing local and global elements in stimuli. Simultanagnosia, as discussed by Baugh et al. (2016), refers to a condition where patients have difficulty perceiving the overall meaning or configuration of complex stimuli. However, they typically retain the ability to perceive isolated elements or details within the stimulus. In integrative agnosia, patients struggle with integrating different elements of a stimulus, resulting in difficulties in recognition. In both cases, the sensory input itself is not degraded, and the impairments are more related to the integration and overall perception of the stimuli.
In contrast, VFA, as proposed here, involves a problem in accurately processing both the local and global elements in stimuli, potentially due to their degradation in quality. In this regard, Nielsen's (1936) key criterion for defining the agnosia is apt: "In this difficulty, the patient has lost sense of direction of lines so that he fails to recognise objects because of their distortion." (pg. 52). In the context of other models of object recognition, such as Marr's model (1982) (as discussed by Warrington and James (1988) with regards to apperceptive agnosia, and Okuda et al. (1996) with regards to patient KK), it might be that VFA patients are only processing visual information up to the level of a 2½-D sketch, where the visible surfaces and contours of discontinuities are present, but they fail to derive a 3-D model of the object in which the geometry (axes and volumetric properties) is fully specified. Indeed, the subjective experiences reported in VFA patients seem to point to a warped phenomenology in structure. In the real world, it is conceivable that some VFA patients use visual cues other than lines and edges to aid object recognition as evidenced by their improved performance with real objects.
At a phenomenological level, we have much to learn about how VFA patients perceive the world, presumably because of the challenges measuring subjective experiences. Nonetheless, Vecera and Gilds (1997) have considered this question. They applied Nagel's (1974) seminal thought experiment of "what is it like to be a bat?" to VFA patients by positing "what is it like to be a patient with apperceptive agnosia?" Nagel's exercise highlights how difficult it is to imagine the subjective sensory awareness that bats may have (if they do experience consciousness) considering they primarily use echolocation to navigate their environment. The exercise is also difficult to apply to VFA patients, but it is perhaps more approachable given they have human brains, albeit damaged ones. Perturb and measure approaches, such as transcranial magnetic stimulation (TMS), could explore this question in neurologically intact individuals (Paus, 2005). Once could use such techniques to transiently disrupt LOC function in both hemispheres and evaluate the consequences that this may have on subjective experiences. As far as we know, this has never been done before.
Nonetheless, previous studies have used TMS to disrupt LOC have demonstrated that it is indeed critical for form processing (Chouinard et al., 2017). But these studies have presented visual stimuli on flatscreens. With good reason, this method of presentation has been the norm in psychophysical experiments for many decades. They have excellent internal validityenabling researchers to tightly control for many confounding variables. However, their ecological validity is limited. One possibility to better understand the role of LOC in form processing in day-to-day perception is to use TMS in conjunction with VR. VR provides the means to better simulate how size and depth operate in the real world by using a virtual three-dimensional space. It reduces (but does not eliminate, as discussed previously) the problem of conflicting depth cues by enabling a more effective simulation of stereopsis and motion parallax while potentially allowing researchers to achieve the same high levels of internal validity as experiments using flatscreens. Using VR may lead to a better understanding of how neurologically intact individuals and VFA patients process and visually perceive the structure of objects in the real world, which has three dimensions.

Orientation and size perception
It is typically assumed that VFA patients have impairments in orientation and size perception (Martinaud, 2017). Thus, one key question we sought to address was: Can patients perceive orientation and size? Not all patients in this review were impaired. Orientation deficits were reported in 9 (43%) cases (RC, DF, PG, SZ, 006, XF, JS, SDV, and SB). We deemed orientation perception to be intact in 6 (29%) cases (HC, SMK, JW, SA, Not Named, and AM) and inconclusive in the remaining 6 (29%) cases. Concerning size processing abilities, 7 (33%) cases displayed evidence of a deficit (DF, SMK, PG, AM, XF, JS, and SDV) while 5 (24%) did not (Mr S, Mr X, FWT, SA, and SB). There was a lack of information to conclude whether abilities were intact or not in the remaining 9 (43%) cases. Given this lack of convergence, problems in orientation or size perception are not always comorbid with problems in shape and form perception. In some patients, there is a dissociation.
The reverse dissociation can be seen in other types of visual agnosia. In some cases of orientation agnosia, recognising objects based on their shape or form is intact but discriminating between the same object presented in different orientations is disrupted (Davidoff and Warrington, 1999;Harris et al., 2001;Martinaud et al., 2016). Although rare, a few cases of dysmetropsia have been reported. In this disorder, objects appear either shrunk (miscropsia) or enlarged (macropsia) compared to their actual size. One case study (Frassinetti et al., 1999) was observed following right occipital ischaemic infarction, which resulted in a permanent left miscropsia. The patient consistently judged the size of objects presented in the left hemifield as smaller than those presented in the right hemifield. In contrast, the patient performed flawlessly when naming objects, faces, and colours. Taken together, orientation and size can be processed independently of formsuggesting that different neural substrates might underlie their perceptual analysis.

Colour perception
Only five patients (24%) demonstrated deficits in colour processing. RC showed impairments when making red/blue judgments. FWT performed poorly at bedside testing where he failed to point to, name, and match items of a specific colour. KK was initially poor at identifying colours, achieving 25% accuracy, when she was first tested. Her performance improved to 60% accuracy at a follow-up testing session. "Not named" and SB performed poorly on several tests of colour vision and were reported to have achromatopsia. In the remaining patients, we deemed colour vision as either normal or inconclusive. Most patients were reported to have normal colour perception as inferred by their performance on the Ishihara Colour Plates, the Farnsworth-Munsell 100-Hue Discrimination Test, and naming visually presented colours.
There is substantial evidence to suggest that form and colour are processed independently from each other. This evidence comes from behavioural studies using the Garner Interference paradigm (Cant et al., 2008;Algom and Fitousi, 2016), electrophysiological studies in non-human primates (Zeki, 1993), and fMRI experiments in humans (Cant and Goodale, 2007). The latter two have identified the involvement of specific brain regions in processing form and colour, namely the LOC for form processing and area V4 for colour processing. In the case of VFA, it is possible that the impairment in colour vision is observed only in a subset of patients whose brain damage extends to regions such as V4 that are crucial for colour processing. This could explain why some VFA patients show deficits in colour perception while others do not. Overall, these findings support the notion that form and colour processing operate independently from each other.
In the seminal study by Milner and Heywood (1989); Milner et al. (1991)on patient DF, a dissociation between preserved colour discrimination and impaired brightness discrimination was observed. This contrasts with findings in achromatopsic patients, where the opposite pattern is typically observed. These results suggest the presence of separate processing channels for colour and brightness discrimination, further supporting the potential independence of colour processing from other aspects of perceptual processing. However, it is important to note that in certain circumstances, colour has been shown to be diagnostically relevant in object recognition. Research has demonstrated that colour can facilitate the recognition of objects (Chouinard and Goodale, 2012;Bramão et al., 2011). This may partly explain the improved performance observed in some patients when identifying solid objects that provide both colour and texture cues. Nevertheless, as mentioned earlier, further research is necessary to gain a better understanding of the factors contributing to the real object advantage observed in some patients. Investigating the underlying mechanisms could help elucidate why and how some patients display this modest advantage. Understanding these processes in patients with VFA and related conditions requires further investigation. Benson and Greenberg (1969) and Efron (1969) emphasised how motion was important for Mr S' residual abilities in perception. He could point to an object in front of him only if it moved. In addition, he could recognise orientation and size with the assistance of motion cues. For example, when shown two discs of light which differed in hue, luminance, and size, he could readily point to the deeper red, brighter, or larger one but only if the two objects were moving. Moreover, when asked if two black sticks of tape on white paper shared the same or different orientation, he carefully followed the contours of each one by moving his head and frequently gave correct answers using this compensatory strategy. He could no longer perform this task when his head remained stationary.

Motion perception
As mentioned in the Introduction, preserved motion perception is typically considered as a feature of VFA. Upon reviewing the case studies, this appears to be largely accurate, although there are some caveats. Although explicit deficits in this domain were not found in any of the reported cases, it is important to note that most patients were not specifically tested on motion perception. Only 6/21 (29%) cases were reported to have preserved motion perception when tested. For the remaining cases, the information was inconclusive or not mentioned. Therefore, it is challenging to draw firm conclusions about the prevalence of preserved motion perception in VFA patients. It is possible that the statistic of approximately 30% preservation may be underestimated due to the lack of explicit testing in many cases and thus some incorporation of such testing in the future may be useful. The cases with preserved motion perception were somewhat varied (Mr S, Mr X, SF, SB, XF, and SDV). Mr S and Mr X both relied heavily on motion for accurate perceptual reporting. For example, Mr S was only able to make judgements about some items if they were drawn in front of him, and Mr X was only successful with recognising some stimuli if he could trace images with his head and hand. Similarly, Mr X, SF, XF, and SDV could judge the direction of moving stimuli. The most in-depth patient tested on motion was SB, whose motion processing was deemed largely normal with an extensive battery (opto-kinetic nystagmus, moving dots displays, moving sine wave gratings), and, in everyday settings, SB reported no difficulty in estimating the speed of walking people or moving cars.
This result is remarkable given the proximity of area V5 (MT) to LOC. No patients aside from SB were reported to have lesions to MT. Interestingly, SB was the patient who had the most extensive motion perception testing, where it was found to be largely normal. One possible explanation for this preserved motion processing in SB despite a notable MT lesion was that this damage was confined to the right hemisphere. As in the case of 006 who still perceived some line drawings successfully potentially due to his unilateral brain damage, it could be the case that the MT in the intact hemisphere was processing motion successfully in SB. Given MT has long been found to be active during motion perception tasks (Zeki, 2015), it is logical that motion processing would be preserved in patients with an intact MT in either hemisphere.

Copying figures
Lissauer (1890) posited that the ability to copy figures is a key distinction point for differentiating between apperceptive and associative agnosia, with the former not being able to do it. This perspective has persisted today with many researchers still stating that VFA patients cannot copy figures and suggesting that this inability should remain a diagnostic feature (Heider, 2000;Farah, 2004;Behrmann and Nishimura, 2010;Martinaud, 2017). Therefore, we asked: Can VFA patients trace or copy figures?
Upon reviewing the case studies for this paper, nine (43%) patients (HC, FWT, SA, AM, KK, NN, JS, SDV, and SB) were able to copy figures. Another case (Mr X) could draw objects comprised of basic geometric shapes when allowed to use his tracing technique. Thus, the inability to copy figures is not consistently present in all patients with VFA. Nearly half can do this. In 6/9 patients who were able to produce some copies, these were limited to basic shapes and copying more complex forms were severely disturbed. Only 'Not Named' could copy some line drawings of objects, although this was a long and laborious process. For example, he took 8 min to produce a copy of a comb. On the other hand, SB and AM made some errors with copying line drawings of objects yet produced quite good copies of the Rey-Osterrieth figure. Thus, Lissauer (1890), distinction between apperceptive and associative agnosia as defined by the absence or presence of abilities to copy figures, respectively, seems out of date. We suggest that this sharp distinction between being able to copy or not copy be updated to specify that there is an inability to copy complex, sophisticated figures in most instances. This distinction strengthens our conclusions that there exists a distinction between basic shape and more complex form.

Imagery
Imagery, like motion, is another feature typically cited as being preserved in cases of VFA. Upon review, this assertion is somewhat unclear, mainly owing to most cases not being explicitly tested on this ability. Three cases had notable deficits in imagery. HC reported diminished visual components in her dreams, Mr X was reported to be deficient without formal testing mentioned, and Mr S claimed to have no dreams or dreams devoid of visual content. However, for most other cases (10/21), the information on imagery was inconclusive or not mentioned, while eight cases (DF, FWT, KK, JW, SZ, SF, SB, and SDV) were reported to have preserved imagery. Testing for imagery was variedin cases with preserved imagery, this was sometimes inferred by their drawing objects from memory, describing objects' physical properties from memory, or mentally comparing different objects according to some physical property to differentiate them. For instance, DF was reported to produce drawings from memory. While SB was normal on several tests including mentally comparing the size of different objects, and mentally comparing a series of items and finding a similarity between them based on some physical property (e.g., global shape). Most cases being inconclusive, along with 3 having notable deficits in imagery, is our main reason behind not concluding that it is typically intact. Incorporating comprehensive testing of imagery is necessary before firm conclusion can be made.
Indeed, the mechanism of imagery is a topic of some controversy. Specifically, the format underlying representations that are engaged and required for imagery has been a point of contention for over 50 years, which continues today. In this debate, Kosslyn et al. (2001) argued that imagery engages the same pictorial representations that are engaged when we experience sight. He argued that the same areas in the brain that are engaged in vision are also engaged in imagery. Pylyshyn (1973) had a different point of view. He argued that imagery is language based, engages symbolic as opposed to pictorial representations, and is processed outside of the visual system.
Recently, this debate has resurfaced in the context of research related to aphantasia (Pearson, 2019). By definition, a person with aphantasia lacks the ability to perceptually imagine visual pictures. However, like many of the patients with VFA that we reviewed, aphants can perform imagery tasksundermining Kosslyn's original views as to what imagery is and what is required to perform imagery tasks. There is growing evidence that early visual areas may actually not be required for imagery. It is of note that VFA patient JW had damage to V1 yet retained imagery. Indeed, this perspective concerning the importance of early visual areas to imagery has been further challenged by Bartolomeo and colleagues (e.g., Bartolomeo et al., 2020). In a comprehensive review of visual agnosia and imagery, Bartolomeo (2021) notes some patients with acquired aphantasia following more anterior lesions in the temporal lobe show deficits of imagery despite intact early visual areas.

Reading
Paradoxically, when you consider that reading might depend on shape and form processing, there have been reports of VFA patients who have retained some reading abilities. For example, HC could write but with many errors (Adler, 1944;Sparr et al., 1991). Counterintuitively, she was better able to read long words compared to shorter ones. This alexia was reported by both Adler (1944) and Sparr et al. (1991) and seemed unchanged between the two series of testing sessions. At the single word level, she regularly misidentified individual letters. While she would read 'front' as 'first' and 'soapy' as 'sorry', she could more accurately read and comprehend longer words such as 'jeopardise', 'physician', and 'sapphire'. Occasionally, when she had difficulty discerning a word, her performance would improve if she used her index finger to trace one or more letters. Likewise, Landis et al. (1982) reported that Mr X was able to read but that this ability was contingent on tracing the letters with his finger. To aid recognition, he developed his own alphabetical code for individual letters (e.g., a 'P' is "a pole and a basketball hoop") and was much quicker at identifying words written in a font saliently abiding to his code compared to other fonts. In addition, despite not having good visual acuity, KK could also demonstrate some abilities in reading. The fact that she could read yet could not recognise line drawings highlights how deficits in visual form agnosia could be specific to non-lexical shapes. This is a point for consideration in future. In many cases, reading abilities were not assessed due to patients initially failing with individual letter recognition. Perhaps more detailed testing of word reading is warranted given cases like HC who could effectively read longer words but struggled with shorter ones.
Our review highlights that 6/21 (29%) retained some abilities to read. HC and DF were similar. Both had more difficulty reading short words (e.g., go, up) than longer ones (e.g., environment). In addition, font type and font size can also affect how easily words could be read in some cases, such as AM (Hildebrandt et al., 2004). Finally, although no qualitative descriptions are available, SA performed well on formal tests of reading like the Warrington reading task (46/50). DF's reading abilities have been carefully examined by Cavina-Pratesi et al. (2015). The authors of this study provide insight into how it was possible for DF to still read words although she cannot perceive shapes or line drawings. This relates to difference in how the two are processed by the brainmechanisms that are intact in DF for the former but not the latter.
DF could read words when presented in a conventional horizontal but not in an unconventional vertical orientation. In addition, she can recognise full words but not individual letters. As it turns out, the visual word form area (VWFA) activates more strongly when people read full words presented in a conventional horizontal orientation (Dehaene and Cohen, 2011). Cavina-Pratesi et al. (2015) demonstrated with fMRI that DF still has an intact VWFA, which is distinct from LOC. It could be the case that basic visual features of words (i.e., line orientations, relative positions, lengths, line-intersections, etc) are processed in DF's intact early visual areas and her intact VWFA can further process this information for word recognition. In contrast, DF has no LOC in either hemisphere to further process visual information for the purposes of recognising non-word shapes.

Recovery in vision
Another point to discuss is the presence of recovery over time. This is not something that is typically considered. It was interesting to observe that a minority of cases were reported to be initially blind (HC, Mr S, DF, FWT, PG, SZ, SB, and SDV). The reason for this is unclear given there is no shared aetiology, although HC, Mr S, and DF all suffered carbon monoxide toxicity. Other kinds of improvement have been observed too. Most cases were only seen on one occasion, but there were a few cases of repeated testing that demonstrate that patients with VFA can show various degrees of recovery. HC was first seen by Adler in 1944. Her impairments then were severe. At a follow-up session nearly 50 years later, Sparr et al. (1991) observed significant improvements. Her impairments only became evident on highly specific tasks (e.g., identifying Poppelreuter figures). FWT was seen shortly after his trauma and nine months later. Some improvements were observed at the second testing session, such as copying figures (3/9 correct from 0/9 correct) and pointing to objects (20/40 correct from 0/30 correct). On the other hand, DF displayed more consistency in her impairments since the late 1980s (Cavina-Pratesi et al., 2015). The reasons for these differences in recovery are unclear. The recovery in FWT could be spontaneous, i.e., abilities return as the brain finds new ways to perform them or naturally repairs some of the damage it sustained. The recovery in HC could partly be spontaneous and partly the result of learning compensatory strategies. In the case of DF, it could be the case that her damage implicated areas such as LOC that are more critical in shape and form processing, enabling her less opportunities to develop strategies to compensate for her deficits.

Recommendations
Based on our review, we have three methodological recommendations for researchers who plan to examine a VFA patient. First, to allow for comparisons across studies, researchers could consider using wellestablished perceptual tests, at least to establish the presence of VFA. For example, some combination of the Snodgrass and Vanderwart images or the BORB for line drawings, along with the Efron task and a simple matching/copying task for shapes, could be considered to verify the presence of difficulties in perceiving shape and form. Second, more detailed structural and functional data on their patient's brain over a period would be informative. Many patients in this review were seen before the emergence of the neuroimaging techniques that we have readily accessible today. DF is the only patient whose brain has been extensively examined over a period with decent resolution imaging. We would have a clearer picture of what brain damage causes VFA and how it progresses over time if more patients were examined in this way. Third, we recommend more thorough testing and disclosure of basic visual function, such as visual acuity and visual field perimetry. We suspect that some previous claims of intact basic visual function or defective basic visual function not accounting for VFA may not be entirely accurate. In some cases, defective basic visual function could have accounted for the agnosia, which is not necessarily problematic.

Conclusions
The findings of the present review have implications for understanding the localisation of brain function. The observation that one site of damage is not common to all patients beyond damage to the occipital lobe provides contradictory evidence for a model of the visual system that exclusively prescribes specific functions to discrete brain areas (e.g., Broca's area being exclusively responsible for language production). Instead, it favours a more hybrid model that considers an associationist view of brain organisation (Weis et al., 2019;Deacon, 2018). Briefly, associationism views the brain as organised in parallel distributed networks around cortical epicentres. In the intact brain, a particular function can be relatively well-localised to a specific area or a set of specific areas. However, a lesion anywhere in the brain could result in a partial dysfunction of other regions that would otherwise be normally connected to the lesioned area. Thus, damage in areas that normally operate with LOC during shape and form processing could equally result in VFA symptoms. Ultimately, more detailed structural and functional data are needed from VFA patients to understand more precisely what neural substrates might be critical.
In sum, VFA is a rare perceptual disorder with a variety of aetiological causes, including carbon monoxide toxicity, stroke, and mercury poisoning. Some clear consistencies exist between patients, such as damage to the occipital lobes in the two hemispheres, impairments recognising line drawings, and intact colour vision. However, as the reader can appreciate, there is no perfect overlap in symptomatology between cases either. As aptly stated by Adler (1944, pg., 243), "the complexity of the process of optic recognition is such that no two patients suffering from the disorder called visual agnosia have identical derangements in function." In our review, we clarify the important distinction between shape and form, the most common perceptual deficits in VFA patients, and note key areas of divergence. Doing this has put into question some widely held diagnostic features, such as intact basic visual function and an inability to copy figures.

Data availability
No data was used for the research described in the article.