Journal Logo


Pure Alexia

A Combined First-Person Account and Neuropsychological Investigation

Hansen, Klaus Cand. Scient.*; Starrfelt, Randi PhD

Author Information
Cognitive and Behavioral Neurology: December 2019 - Volume 32 - Issue 4 - p 268-277
doi: 10.1097/WNN.0000000000000214


IBOS: Institut for Blinde og Svagtseende (Institute for the Blind and Partially Sighted)

RT: reaction time

WLE: word-length effect

This paper is a first-person account of pure alexia as experienced by the first author (K.H.), followed by a description of the remission process from severe reading problems acutely following stroke to mild pure alexia today. This experience, coupled with K.H.’s knowledge of computer models of vision, led to discussions between K.H. and the second author (R.S.) about the nature of the reading system and what K.H.’s reading system must look like in order to explain his reading deficits. On this basis, K.H. has sketched a model of reading within which his deficits, and also potential compensating strategies, may be explained—which we aim to disseminate in full in the future. Here, we present K.H.’s firsthand account of his reading deficit, as well as his remission from the acute stage poststroke and throughout the next 7 years until today, along with background neuropsychological data and reading measures taken at four time points during K.H.’s recovery. In addition, data from K.H.’s participation in studies relating to reading and visual recognition are included.


Pure alexia, or alexia without agraphia, is a deficit in reading skills that is acquired after injury (often stroke) to the left posterior ventral cortex. As a result of this deficit, reading may be completely abolished, as was the case with Dejerine’s original patient (Dejerine, 1892), particularly in the acute phase. More commonly, patients retain the ability to identify single letters (albeit slowly) and read words letter-by-letter. In the early phase of remission, this strategy might be overt, with actual naming of the individual letters, but many patients remit to the point where their serial letter identification is inferred from the measurement of word-length effects (WLEs; a monotonic increase in reaction times [RTs] with word length) in their single-word reading (Starrfelt and Shallice, 2014). Most patients with pure alexia also have visual field defects affecting the entire right visual hemifield, or the upper right quadrant.

Several different explanations for pure alexia have been proposed over the years since Dejerine’s (1892) original suggestion that pure alexia is a disconnection syndrome. The fact that writing is preserved indicates that a perceptual rather than a language deficit is at the core of pure alexia; however, the nature of this perceptual deficit has remained elusive. In the neuropsychological literature, the focus has mainly been on explaining the WLE (Barton et al, 2014), which is thought to reflect a serial encoding strategy that compensates for a breakdown of parallel letter processing (Shallice, 2014). The main debate concerns whether the core deficit in pure alexia is on a cognitive or cerebral level solely concerned with processing alphabetic characters, that is, a word-form level or area (Cohen et al, 2004; Warrington and Shallice, 1980), or if it may be explained by a more general visual deficit with a disproportional effect on word recognition (Behrmann and Plaut, 2013; Starrfelt et al, 2009). In pure alexia, the lesion typically includes the left (mid-) fusiform gyrus, either directly involving the putative visual word-form area or disconnecting this area from visual input (Dehaene and Cohen, 2011; Leff et al, 2006).

Regarding the pattern of remission from pure alexia, there are a few case studies documenting such remission using behavioral (Behrmann et al, 1990) and functional imaging measures (Cohen et al, 2016). These reports converge on the conclusion that patients do not recover normal reading even years after their injury; moreover, Cohen et al (2016) suggested that the anatomy underlying letter-by-letter reading skills is likely to include general-purpose visual areas that are not capable of fast, parallel letter processing.


The following text includes K.H.’s personal account of his background and life before his stroke, immediately after the stroke, and the later development of his symptoms. Each section also contains relevant information from K.H.’s medical files as well as cognitive assessments from studies that K.H. participated in after his stroke.

Background Information

K.H.’s Personal Account

I was born in 1947 and have a degree in computer science and mathematics from the University of Copenhagen (equivalent to a PhD in the United States). I was a systems programmer and software designer in a private company, developing hardware and software for business and networks for 7 years. Following that, I was an associate professor at the computer science department at the University of Copenhagen for 30 years, teaching a broad selection of topics, including programming, operating systems, and pattern recognition.

I have always been good at reading, writing, and speaking Danish, experiencing no trouble with spelling or pronunciation. I was permitted to borrow books at the school library after a year in school and have been reading constantly since, often reading two books at a time in addition to books relating to my education. As an adult, I have read much both professionally and for pleasure, and believe that I have been a very fast reader (judging, for example, by the time I would take to read papers distributed at meetings compared to the other people reading the same text). I was also very quick at scanning texts for keywords and skimming to judge the overall content of a text.

After 6 years of learning English, starting at age 11, I had a reasonable fluency in reading and speaking English and, to a lesser degree, writing English. Later, when servicing customers in Europe and participating in international meetings, my English became a second language, and I read parts of the standard literature in the English language. I learned German for 3 years in school, starting at age 12. I had 1 year of Latin at age 14, and 3 years of French, starting at age 15. I have some understanding of these three languages, although I do not speak or write them. I also understand the Nordic languages but do not speak or write them. In college, I studied classical Greek for 2 months but had to stop for practical reasons.

At the age of 7 years, I learned basic musical notation. I learned to read more advanced musical notation at around age 10. I have been practicing music and notation reading for approximately 55 years and, before my stroke, I was an amateur viola player with relatively high skills. Thus, I have excellent knowledge of advanced musical notation and classical music. I learned mathematical notation at around the age of 14 and advanced mathematical notation from 16 to 22 years of age. I also have knowledge of the main programming languages, including Algol, Pascal, C, C++, and several machine languages, as well as knowledge of advanced mathematical topics, including information theory, differential and integral geometry, and statistics. I have always had an excellent sense of locations and directions in three-dimensional space, making it easy, for example, to plan a shopping route in large shopping centers.

K.H.’s Premorbid Reading Skills

In an attempt to quantify K.H.’s premorbid reading ability and experience for comparison purposes, he was asked to complete the Adult Reading History Questionnaire (Lefly and Pennington, 2000). This questionnaire is designed to identify developmental reading disorders retrospectively. It contains 23 questions with responses on a Likert scale, ranging from 0 (no reading problem/much reading experience) to 4 (reading problem/little reading experience). The questionnaire also contains three questions used for “informational purposes,” which we (R.S.’ research group) did not include. We have adapted this scale for use with stroke patients, adding the phrase “before your stroke” to all relevant questions so as to ensure that patients report their reading skills from the time before their injury. Four questions are left out in this version; three that pertain to verbal memory and not reading experience (questions 16–18), and one specifically asking about reading Sunday newspapers (question 23). This adapted version thus includes 19 questions. The maximum possible score in this adapted version is 76; dividing a patient’s score by 76 yields a score equivalent to the standard Adult Reading History Questionnaire score, where a score >0.30 is considered indicative of a history of a reading disorder (Lefly and Pennington, 2000).

We administered the questionnaire to K.H. 6½ years poststroke. All questions and response alternatives were read out loud by R.S., and K.H. was asked to answer them orally. K.H. scored 1 out of 76 on this questionnaire (score of 0.013), indicating that he had not experienced any reading difficulties before his stroke but rather learned to read early and typically performed above the average reading level in school. K.H.’s score also indicated that he could read fluently before the stroke and that he had read both professionally and for recreation throughout his life.

The Stroke

K.H.’s Personal Account

After a healthy life without any illnesses, I had a minor stroke at the age of 64. A few days before the stroke, I experienced strong zig-zag patterns in the right part of the visual field, so I went to the emergency room of my local hospital, where it was discovered that I had a highly elevated blood pressure. Aside from these visual patterns, which had occurred a couple of times in the years preceding the stroke, the only symptoms of high blood pressure I had experienced were occasional nosebleeds. I did not have headaches in the period before the stroke. In fact, I never had migraines and very seldom had headaches and never experienced hallucinations. The following account is based on my memory of the events following my stroke.

Three days after the zig-zag episode, I had a stroke that caused bilateral loss of the right half of my vision. The stroke occurred at night while I was asleep, and I was admitted to the same hospital in the morning. In the following days, I experienced an inability to read, but as I recall it, the hospital did not confirm this inability during my 10-day stay. I did not experience difficulties with spoken language or writing. One problem I did experience immediately after the stroke was difficulties in determining, for example, which letter in the group “b d p q” I was actually trying to read. This was also the case for “R P,” but I experienced no problems with “a e s A H K S” or the digits “0123456789.” Another potentially related phenomenon was that, when I was tired, I had difficulty with determining right from left, up from down, and front from back. This difficulty resulted in my saying, for example, “turn left” when my actual intention was to say “right,” or in picking up a pencil and realizing that I was trying to use the wrong end and having to turn it around. The up/down direction seemed less prone to errors than right/left. During the first months after the stroke, I also had problems orienting myself in space. In the first few months after the stroke, these problems were so severe that I was unable to access my inner three-dimensional map or recognize where I was. This problem has since virtually disappeared.

Following the stroke, I was generally very tired, and all activities felt demanding. I needed to rest much more than I was used to.

The stroke placed me in a situation in which I had no control over what happened inside and outside me. Emotionally, I felt it was a matter of survival, especially because a function essential to me (reading) was in danger of being lost. Immediately following the stroke, I had an intense feeling of lack of control over the situation and the future. I felt that the things that happened had directions and results I could not predict. I did not experience depression at this point, possibly due to my sessions with a psychologist a couple of years before, and partly because it seemed to matter what I did to better my situation. Speaking to doctors and other specialists made me feel moderately optimistic about my condition.

I did not focus on or share many of the emotions I had about the stroke. Since my early teenage years, I have kept most of my feelings and opinions inside, carefully controlling what other people could see. In 2008, I received psychological help because of a depression that caused a loss of interest in my job. The conclusion from seven sessions was that I had no hidden trauma, but my main problem was my wish to perform 100%. I had developed a method for dealing with emotional stress by making an emotional package of facts, events, and feelings that I could store in my memory and then ignore, later replaying it in my head with a fair amount of details (this technique is quite useful when debugging software or playing music). I used this method following my stroke, too, which allowed me to focus on the more concrete, practical aspects of recovery. The first days at the hospital, for example, I focused on getting the basic functions working again; that is, walking and eating. My mood at that time was uncontrollable, and I cried at several occasions, which is something I very seldom do under normal circumstances. In the first couple of years, this inclination to weep did not go away, and still some music touches me emotionally and causes a stronger reaction than it would have before the stroke.

From the Medical Files

The following text is based on information in the hospital medical files (provided by K.H.). K.H. was admitted to the hospital following an experience with seeing zig-zag patterns. Severe hypertension was noted, and treatment was prescribed, after which K.H. was discharged. This incident was followed 3 days later by a stroke that occurred during sleep. K.H.’s main acute symptoms, as noted in the medical files, were intense left-sided headache; a dense, right-sided hemianopia; and a feeling of reduced sensation in the right hand and foot. Upon admittance to the hospital, a dense right-sided hemianopia, as well as mispointing with the right hand in the finger-nose-finger test, was noted by the neurologist on call. No other neurologic symptoms were recorded at that time, but a high blood pressure was registered. An acute CT scan (on the day of admittance) revealed a left occipital stroke. During the 10-day hospital stay, a complete right hemianopia without macular sparing was diagnosed based on confrontation testing by a neurologist. Severe reading difficulties were noted during a neurologic exam, and the files state that the patient could only read by identifying single letters and putting them together.

Five days after admission, K.H. was referred to a speech-language pathologist for a language evaluation because of his reading problems. This evaluation concluded that impressive and expressive language and writing were unimpaired. Slight naming problems were noted, but were suggested to be habitual rather than a result of the stroke. Reading problems were noted, and these were suggested to be caused by visual problems rather than linguistic problems with reading. The speech-language pathologist’s conclusion was that the patient did not have alexia. K.H. was discharged from the hospital 11 days after admission and was referred to the Institute for the Blind and Partially Sighted (Institut for Blinde og Svagtseende, IBOS) for rehabilitation focusing on compensation for his hemianopia. He was referred to a clinical neuropsychologist for a follow-up 2½ months later. However, there is no note in the hospital files about such an assessment.

According to the file from IBOS, K.H. received extensive training of visual search and scanning (with NeuroVision Technology; George et al, 2011) related to his hemianopia, as well as mobility training to help him compensate for the hemianopia when moving around, for example, in traffic. He received no formal reading therapy while at IBOS.

Reading Recovery

K.H.’s Personal Account

This section describes how I advanced to my current state, where I can read longer articles in newspapers and magazines and have finished two 450-page books in English as well as one Danish book containing 622 pages of small print. My reading speed has now (6½ years poststroke) improved as far as to allow me to catch approximately 80% of the subtitles in movies.

The hospital stay focused on my somatic problems rather than the reading problem. After leaving the hospital, a rehabilitation program was planned by IBOS. This program focused on training attention and scanning the surroundings to compensate for the loss of parts of the viewing field. In connection with this, I took a reading test. My own observation was that in addition to my problems with reading words and text, I had problems reading basic musical notation, but my mathematical notation reading ability was unaffected.

The following account is based on information in the file from IBOS as well as my memory of the events. The result of the reading test at IBOS at 6 weeks after the stroke was that the functioning of the eyes and the basic part of the processing of visual stimuli were intact. The test showed that I read single letters correctly for small letters (24/24 correct), but I had a significant problem reading capital letters (19/24 correct). Reading aloud a collection of words of three to nine characters in length resulted in an error rate of 4 out of 60 words, but for words of five or more characters in length, my reading was very slow. However, I was able to read 12 out of 12 numbers consisting of one to five digits correctly. The general conclusion was that regaining my reading ability was mostly a matter of training, but no formal rehabilitation program was set in place. Therefore, I began my own training program.

The first part of my self-initiated training centered on recognizing letters and common sequences occurring in printed text. My program consisted of reading a newspaper and solving Sudoku and crossword puzzles, spending typically 1–2 hours on this each day. Early in this training (3–4 months poststroke), I experienced a new problem: When attempting to read a sentence, some words clearly did not make any sense in the context, and re-reading the text showed that I had been guessing what words came next rather than reading them. I believe that this was part of my reading strategy before the stroke: reading fast and sometimes guessing the text from cues that I had not yet consciously perceived. As this method clearly did not work anymore, I spent a period of several weeks to unlearn it. During this period, I started paying attention to each letter in each word to ensure that I identified them all correctly. I also went back to the beginning of the sentence whenever I encountered a word that did not make sense in the context. My reading speed at this point (4–5 months poststroke) was <10 words/minute, each word being read slowly so that errors were avoided. The estimate of words per minute was based on timing the reading of a newspaper article and counting the number of words. The estimated number of errors was based on the number of errors I detected, for example, when words did not fit with the context.

After a year of training, my main problems were speed and getting the last syllables of long words right. My reading speed at this point (12–14 months poststroke) was approximately 30 words/minute. A year later (25–28 months poststroke), my reading speed had increased to 70 words/minute, and the number of reading errors had significantly decreased. At the beginning of 2017, about 56 months after the stroke, my reading speed had further increased to 75 words/minute. This makes it possible to read subtitles in movies with a moderate speed of speech. In addition to this, I am now able to scan texts for keywords, in contrast to the situation at 3–4 months poststroke, when I had to scan text word-by-word and line-by-line. I feel I make significantly fewer errors in reading now, even in subtitles or when reading books at a (for me) high speed, and that my self-initiated training in focusing on the letters in the word and avoiding guessing has paid off.

During 2016, about 4 years after the stroke, I experienced a plateau in my reading; there were no speed improvements for some time. This suggested to me that there is a natural limit of the method I now use for reading, corresponding to around six characters per second. During the last formal reading assessment to date (6½ years poststroke), I noticed a kind of competition in my mind when reading the single words presented on the computer screen. It felt like I had access to a fast response that was immediate, and which I sometimes responded with. In other instances, I felt unsure about this response and shifted to a strategy of error checking, which was slower and included a stronger focus on the individual letters in the word. This strategy reminded me of the early phases of my recovery, where I would read a text and understand from the context that I had made a mistake and had to go back in the text. Perhaps I now have a more automatic form of error detection or checking, which signals when I am not certain enough that I have read the correct word.

Reading Musical Notation and Other Symbols

About 6 months after my stroke, I tested my ability to read musical notation. The result was that I had problems translating the notation to finger position and string. In my experience, a trained violinist may read up 70 notes per minute when playing a musical piece for the first time. The fastest pieces of classical music, like the violin solo in the fifth Brandenburger Concerto by J. S. Bach (1, movement), or his Prelude (Fantasia) in A minor, BWV 922, can play at up to 20 notes per second, which is only possible if the music has been learned by heart. Speeds in between call for prelearned sequences, with the written music being used as an orientation tool. The other elements of reading music (eg, judging pitch and temporal duration) seemed to me to be intact. When reading advanced mathematical notation (learned at the age of 16 and later) and computer programming language constructs, I experienced no problems; however, I did not test the various elements (eg, design, writing code, debugging it, reading comments) in detail.

My reading of musical notation did improve, along with my reading of words and letters—at least, that is my feeling. Although I have not practiced to the degree I did before my stroke, I have tried playing my viola several times since the stroke. In the year after the stroke, I had difficulties hitting only one string, using the right amount of pressure on the bow, and recognizing the alto-key used when reading music written for the viola. Now, most of my motor skills have returned and I can read music, although not at the same speed as before the stroke. Similarly to my experience of a plateau, or bottleneck, in reading words, I experienced a limitation in decoding musical notation that I did not have before my stroke; now, I can read only a couple of notes per second.

My Experience of the Blind Field

The stroke caused the right half of my visual field to disappear completely from the conscious part of my vision, leaving an empty place unlike what normally is outside the field of vision. To be clear, in the first month, I experienced it as a black hole rather than just being unconscious of it. After a short period of time, I experienced a feeling of something that was not visible being there. At the same time, I began having problems focusing and in the fusion or integration of the visual fields of the left and right eye. From time to time, I still experience this problem when I am tired. About a month after the stroke, I stopped seeing the black hole, and my intact visual field became my whole field of vision; that is, I now perceived what was in the blind field as everything else outside my visual field (eg, what is behind me): I know it is there, but I do not see it.

Neuropsychological and Medical Data

Neuropsychological Screening 1 Year Poststroke

In May 2013, 1 year poststroke, K.H. contacted R.S. about his reading problems, having been inspired by media coverage of a symposium on pure alexia. A month later, K.H. underwent a neuropsychological screening, which revealed only minor deficits (Table 1) compared to a control group from an ongoing project (25 participants close in age but slightly less educated). As we had no formal information about K.H.’s visual field defect at this time, R.S. performed an in-house computerized perimetry (Petersen et al, 2016), which revealed a dense right hemianopia affecting the entire hemifield but seemingly sparing the central 1 to 2 degrees (this is uncertain, as it is at the limits of the test resolution).

K.H.’s Test Results 1 Year Poststroke

The most notable aspect of the test results from this time point is that K.H.’s RTs in naming single letters, digits, and objects were within the control range, whereas his single-word reading RTs and WLEs were significantly outside the control range and on a level compatible with a diagnosis of pure alexia. K.H.’s writing, as tested using the Writing subtest of the Western Aphasia Battery (Kertesz, 1982), was normal at this time. Additional evidence of K.H.’s intact writing is his continuing email correspondence with R.S. as well as the fact that K.H. wrote the first draft of the first-person account in this paper.

Lesion Localization

A high-resolution MRI performed 15 months poststroke (in the context of a research project) showed a lesion in K.H.’s posterior left cerebral hemisphere, affecting the mesial part of the left temporal and occipital lobes (Figure 1; additional images showing the extent of the lesion are provided as Supplemental Digital Content, Anteriorly, the lesion involved major parts of the left parahippocampal and lingual gyri; however, the most anterior part of these structures was spared. The body and tail of the hippocampus were also affected, resulting in discrete widening of the left choroidal fissure. Posteriorly, the lesion almost completely affected the lingual and fusiform gyri, as well as the cuneus. Mesially, the cortex was partially spared. Superiorly, the lesion extended to the calcarine sulcus. There was an additional lesion in the posterior part of the left thalamus as well as several small dot-like lesions centrally in the splenium of the corpus callosum. The MRI findings were published in Petersen et al (2016) and Asperud et al (2019).

K.H.’s structural MRI 15 months poststroke. Images depicted in radiological convention (left hemisphere depicted on the right).

Visual Field

A right hemianopia was revealed at hospital admission, and this field defect persists, although K.H. sometimes experiences a sense of vision in the outer boundaries of his peripheral right field. Repeated perimetric evaluations from an ophthalmologist show a dense right-sided homonymous hemianopia with no significant change since the first evaluation, 1 year poststroke. Figure 2 shows the most recent evaluation of K.H.’s visual field, measured more than 6 years poststroke, and, for comparison, 3½ years poststroke. Both show practically no preserved vision within the 30-degree right field.

K.H.’s visual field. Top panel shows perimetry performed 3½ years poststroke. Bottom panel shows perimetry performed almost 6½ years poststroke. Both show a dense right hemianopia, which may not be entirely homonymous (slightly more preserved central vision on the left eye). Over the years, there has been little change between measurements of the visual field, although the noncentral field in the right eye does appear to be expanding. (Figure 2 can be viewed in color online at

Clinical Testing and Progress Details

Following the initial neuropsychological testing reported in Table 1, K.H. has participated in several other tests and projects at the psychology department at the University of Copenhagen (Asperud et al, 2019; Petersen et al, 2016; Sand et al, 2018). A summary of the findings is presented in the following paragraphs.

Reading Data From K.H.’s Recovery

K.H. has been tested using different sets of regular words for the measurement of RTs and WLEs; this was carried out as part of different studies, at four different time points (see Table 2 for accuracy data and RTs at different time points). Overall, K.H. has made few reading errors (2013: 2/150 errors, 2015: 1/52 errors; 2016: 0/75 errors; 2019: 5/150 errors), and his RTs and WLEs have decreased over time. For comparison purposes, mean control RTs in these reading tests are typically around 500 to 600 milliseconds, and WLEs for controls are in the range of 0 to 30 milliseconds per letter.

K.H.’s Reading Data Throughout the Years Poststroke

The reading test used in the initial evaluation listed in Table 1 was a word list consisting of 150 nouns of five to seven letters in length (50 for each word length). K.H.’s mean RT in this test 1 year poststroke was 1598 milliseconds (SD=525), with a WLE of 252 milliseconds per letter. In a study conducted 2 years poststroke, K.H. took a shorter test with nouns of three, five, and seven letters. His mean RT on the test at that time was 1132 milliseconds (SD=532), with a WLE of 188 milliseconds per letter (Asperud et al, 2019; Petersen et al, 2016). Four years poststroke, K.H. took a shortened version of the original reading test (25 nouns of five to seven letters in length). At that time, K.H.’s mean RT was 1151 (SD=353), with a WLE of 127 milliseconds per letter (Sand et al, 2018). Most recently, 6½ years poststroke, K.H. took the original reading test again for the purpose of this paper. At this latest assessment, K.H.’s mean RT was 1470 milliseconds (SD=487), with a WLE of 164 milliseconds per letter. Note that the mean RTs are not comparable between tests because they include words of different lengths, but the WLEs should (in theory) be unrelated to the word lengths used to measure them. The WLE, then, was initially on level with that of other patients in the literature with pure alexia (eg, Starrfelt et al, 2009, 2010), whereas in more recent years (2 years poststroke and onward), the WLE fluctuates around the typical cutoff suggested to differentiate between hemianopic and pure alexia (usually ∼150 milliseconds per letter) (Barton et al, 2014; Leff et al, 2001).

Another interesting aspect of K.H.’s reading RTs is the large variability within word lengths, as indicated by the SDs of the mean RTs across the years. This variability is often present in patients with pure alexia (see Leff and Starrfelt, 2014) but is poorly understood. Although some of this variability around the mean is clearly explained by word length, there is also large variability around the means for each word length, as illustrated in Figure 3 for the latest measure of K.H.’s reading RTs. Linguistic variables like frequency and part-of-speech have not been the focus of any of these assessments, all presented words have been nouns with regular spelling; therefore, we cannot judge whether there may be effects of such variables on K.H.’s reading. It is interesting in the current context that some of the RTs are fast (<1000 milliseconds), whereas others are longer than 2000 milliseconds, even for three-letter words. This might relate to K.H.’s feeling that he has two different reading strategies at his disposal; one fast, which involves some guessing, and another, slower route that includes focusing more on the constituent letters of the word.

K.H.’s individual reading times in word reading 6½ years poststroke. Each point is a reading time for a single word, clearly illustrating the large variability in KH’s reading times even within the same word length. Line represents linear regression (the word-length effect). RT=reaction time.

Other Neuropsychological Data

Although most of the studies K.H. has participated in have used computerized, experimental tests, K.H. was also asked to complete a series of subtests using nonoverlapping figures from the Birmingham Object Recognition Battery (Riddoch and Humphreys, 1993; Table 3), as part of the assessment reported in Sand et al (2018). In these tests, K.H.’s accuracy was similar to that of the controls in all subtests (he made one naming error across all subtests; control participants made 0–3). The Birmingham Object Recognition Battery was designed to detect deficits based on errors, but also includes measuring RTs with a stopwatch (time per item=time per page divided by the number of items presented on the page). On this measure, K.H. was significantly slower than the controls for single, double, and triple letters as well as single and double drawings (objects); however, he was not significantly slower for single, double, or triple shapes (Table 3).

KH’s Reading Times (Time Per Item) for the Nonoverlapping Figures Subtests From the Birmingham Object Recognition Battery

Visual Attention and Crowding

Petersen et al (2016) showed that K.H. has intact visual processing speed and visual apprehension span in his intact left visual field when measured with a psychophysical whole report paradigm with brief presentation of several letters at once. In his hemianopic right visual field, K.H. could not identify a single letter. More typically, patients with pure alexia show general reductions in processing speed and apprehension span in both hemifields, which has been suggested to contribute to their reading deficit (Starrfelt et al, 2009, 2010). In comparison, a patient with mild hemianopic alexia showed normal processing speed and apprehension span in the intact visual field (Habekost and Starrfelt, 2006; Starrfelt et al, 2009). It is tempting to speculate that K.H.’s intact processing speed and apprehension span may have contributed to his reading recovery, but this remains a speculation. In a psychophysical, clinically usable test of visual crowding (devised by Pelli et al, 2016), which measures acuity and crowding for letters and digits, K.H.’s acuity and susceptibility to visual crowding in his foveal vision was found to be on level with controls, indicating that crowding cannot explain his reading problem (Sand et al, 2018).

Face Recognition

Based on the hypothesis suggested by Behrmann and Plaut (2014) that patients with pure alexia also show face recognition problems (although milder than in acquired prosopagnosia), K.H. has been tested with several tests assessing face recognition performance. One year poststroke, K.H. recognized 100% (20 out of 20) of famous Danish faces (Jørgensen et al, 2009) and could name 18 of these. This score is within the normal range based both on published normative data and on the scores from a control group (Table 1). Four years poststroke, K.H.’s face recognition abilities were tested further by Sand et al (2018). They reported that K.H. scored outside the control range on Warrington’s Recognition Memory Test (Warrington, 1984) for both words and faces (scoring 30/50 and 33/50, respectively). He also scored outside the control range on the Cambridge Face Memory Test (Duchaine and Nakayama, 2006), although his score was not significantly abnormal based on single-case statistics (he scored 48/72, P=0.061, Crawford and Howell’s (1998)t test, one-tailed).

In a study of word, object, and face recognition using a delayed matching task with words, cars, full faces, and cropped faces, Asperud et al (2019) found that K.H.’s accuracy was within the normal range when matching words, cars, and cropped faces, whereas his RTs were slightly but significantly elevated when matching full faces (K.H. 899 milliseconds, Controls 677 [SD=99]). Note that the word condition in this test consisted of four-letter words, where one of the two middle letters was changed in different-trials. K.H. spontaneously commented after completing this test that he had discovered this and only looked at the two middle letters when solving the task. His within-normal range results, then, were interpreted to reflect his ability for single-letter identification rather than word reading (Asperud et al, 2019).

Given that K.H. has never complained about difficulties recognizing faces following his stroke, his reduced scores in some face tests prompted me (R.S.) to ask him about his face recognition abilities using the face recognition part of the Faces and Emotions Questionnaire (Freeman et al, 2015). This questionnaire consists of a series of statements about everyday face recognition, and participants are asked to report to which degree they agree with the statement. The questionnaire is freely available online ( and has been translated to Danish. There are no published norms for this questionnaire, but we have used the Danish version in several control groups in studies of face recognition. Although the participants in these control groups were much younger than K.H. (M age=35 and 37 years [range=17–56] in the groups reported below), they can serve as an indicator of whether his self-reported face recognition abilities suggest problems in everyday life. In two of these control groups (both reported in Hendel et al, 2019), the mean scores were 16.7 and 21.6, respectively, whereas scores for participants with developmental prosopagnosia are typically >50 (indicating problems with face recognition). K.H.’s score on this questionnaire was 18, indicating that he does not experience difficulties with face recognition in everyday life any more than the control participants, in spite of his low performance on tests of face matching and memory in the laboratory.

K.H. can remember and recognize known faces in tests and everyday life (Famous Faces Test and Faces and Emotions Questionnaire), even though he has slightly elevated RTs in delayed matching of faces (which may be due to a combination of his hemianopia obscuring parts of the face and the strategy he applies) and a deficit in remembering newly encountered faces (Warrington’s Recognition Memory Test and Cambridge Face Memory Test). A tentative explanation for K.H.’s face recognition performance, then, is that it reflects a slight memory deficit, perhaps related to his hippocampal lesion rather than being of perceptual origin. In this context, it is interesting that although K.H. scored within the normal range on the Mini-Mental State Examination (Folstein et al, 1975) in the first assessment (ie, score of 28, 1 year poststroke), he missed two points because he remembered only one of the three words. During a meeting 6 years poststroke, K.H. was asked about his memory and reported that he does not feel that his general memory has changed following his stroke.

Summary of Neuropsychological Data and Experimental Tests

In summary, K.H.’s elevated RTs and WLEs in reading persist more than 6 years poststroke, although neither is as severe as in the first year poststroke. Indeed, in the last few years, K.H.’s WLE has come close to the cutoff between hemianopic and pure alexia (typically ∼150 milliseconds). Interestingly, his RTs in naming of single letters and digits was already within the normal range 1 year poststroke. His writing has been unimpaired throughout and remains so. Few other cognitive or perceptual skills have been affected since he was first tested 1 year poststroke. His accuracy and RTs in picture naming and recognition were in the normal range 1 year poststroke, as was his performance on visual neuropsychological tests, such as the Star Cancellation Test (accuracy) (Wilson et al, 1987), the Overlapping Figures Test (Christensen, 1979), and the Street Completion Test (Gade et al, 1988).

Additional experimental testing has revealed that K.H.’s visual processing speed and apprehension span are within the normal range in the intact left visual hemifield (Petersen et al, 2016). On two measures of foveal visual crowding, he also scored within the normal range (Sand et al, 2018). One exception is his performance on the Warrington Recognition Memory Test, which was clearly impaired 4 years poststroke, as well as the Cambridge Face Memory Test, where his performance was borderline impaired at the same time point. In light of his normal recognition of famous faces 1 year poststroke, as well as his self-report of no difficulties in recognizing faces in everyday life (including in a questionnaire), it is tempting to interpret his performance on the Recognition Memory Test for Faces and Cambridge Face Memory Test as difficulties in recalling newly encountered faces rather than a perceptual or face-recognition deficit; however, this is speculation.


Overall, the measured improvement in K.H.’s single-word RTs is in agreement with his self-report of improved reading at least to some degree. It seems, however, that his perception of his improvement is greater than what has been measured in the laboratory. Part of the reason for this difference is probably that much of the remission occurred in the first year poststroke, before K.H. contacted our laboratory. At present, K.H. could possibly be characterized as having hemianopic rather than pure alexia, if diagnosis was to be based only on his RTs and WLEs. However, his rather long RTs for many words indicate that his reading is impaired by more than his hemianopia, while it seems he also has some responses that are too fast to reflect letter-by-letter reading. K.H.’s description of having two automatic but competing reading strategies may help explain the variability in his RTs. He experiences a fast, automatic response, which sometimes comes with an error signal that prompts him to look more closely at the word. We hope to present K.H.’s explanation for this, as well as how it may be interpreted within standard models of reading, in a future paper describing a model of reading that K.H. has developed based on his reading deficit and remission.

Some Thoughts on the Future

I find the progress I have made in reading over the last years quite satisfying, and I expect that further progress in reading speed and accuracy is still possible. Working with the present paper has made me aware of other problems I have, in particular with concentration, which I feel I have overcome to a large extent during the work done for this paper—which is very satisfying. The lack of vision on my right side is still a problem for me and seems like it will never disappear. In general, though, I am fairly optimistic about the future and am looking forward to working on a follow-up manuscript on a model for reading.


Asperud J, Kühn CD, Gerlach C, et al. 2019. Word recognition and face recognition following posterior cerebral artery stroke: overlapping networks and selective contributions. Vis Cogn. 27:52–65. doi:10.1080/13506285.2019.1599481
Barton JJ, Hanif HM, Eklinder Bjornstrom L, et al. 2014. The word-length effect in reading: a review. Cogn Neuropsychol. 31:378–412. doi:10.1080/02643294.2014.895314
Behrmann M, Black SE, Bub D. 1990. The evolution of pure alexia: a longitudinal study of recovery. Brain Lang. 39:405–427.
Behrmann M, Plaut DC. 2013. Distributed circuits, not circumscribed centers, mediate visual recognition. Trends Cogn Sci. 17:210–219. doi:10.1016/j.tics.2013.03.007
Behrmann M, Plaut DC. 2014. Bilateral hemispheric processing of words and faces: evidence from word impairments in prosopagnosia and face impairments in pure alexia. Cereb Cortex. 24:1102–1118. doi:10.1093/cercor/bhs390
Cohen L, Dehaene S, McCormick S, et al. 2016. Brain mechanisms of recovery from pure alexia: a single case study with multiple longitudinal scans. Neuropsychologia. 91:36–49. doi:10.1016/j.neuropsychologia.2016.07.009
Cohen L, Henry C, Dehaene S, et al. 2004. The pathophysiology of letter-by-letter reading. Neuropsychologia. 42:1768–1780. doi:10.1016/j.neuropsychologia.2004.04.018
Christensen AL. 1979. Luria’s Neuropsychological Investigation. Copenhagen, Denmark: Munksgaard.
Crawford JR, Howell DC. 1998. Comparing an individual’s test score against norms derived from small samples. Clin Neuropsychol. 12:482–486.
Dehaene S, Cohen L. 2011. The unique role of the visual word form area in reading. Trends Cogn Sci. 15:254–262.
Dejerine J. 1892. Contribution a l'étude anatomo-pathologique et clinique des differentes variétés de cécité-verbale [Contribution to the anatomical and clinical study of different kinds of word blindness]. Memoires Societé Biologique. 4:61–90.
Duchaine B, Nakayama K. 2006. The Cambridge Face Memory Test: results for neurologically intact individuals and an investigation of its validity using inverted face stimuli and prosopagnosic participants. Neuropsychologia. 44:576–585.
Freeman P, Palermo R, Brock J. 2015. Faces and Emotions Questionnaire. Version 1. Figshare. doi:10.6084/M9.FIGSHARE.1501441
Folstein MF, Folstein SE, McHugh PR. 1975. “Mini-mental state”. A practical method for grading the cognitive state of patients for the clinician. J Psychiatr Res. 12:189–198.
Gade A, Udesen H, Mortensen EL. 1988. Visual closure—Street Completion Test. Nordisk Psykologi. 40:194–201. doi:10.1080/00291463.1988.10636922
George S, Hayes A, Chen C, et al. 2011. The effect of static scanning and mobility training on mobility in people with hemianopia after stroke: a randomized controlled trial comparing standardized versus non-standardized treatment protocols. BMC Neurol. 11:87. doi:10.1186/1471-2377-11-87
Habekost T, Starrfelt R. 2006. Alexia and quadrant-amblyopia: reading disability after a minor visual field deficit. Neuropsychologia. 44:2465–2476.
Hendel RK, Starrfelt R, Gerlach C. 2019. The good, the bad, and the average: characterizing the relationship between face and object processing across the face recognition spectrum. Neuropsychologia. 124:274–284. doi:10.1016/j.neuropsychologia.2018.11.016
James M, Plant GT, Warrington EK. 2001. Cortical Vision Screening Test (Corvist). London, England: Harcourt Assessment.
    Jørgensen K, Stokholm J, Vogel A. 2009. Kendte ansigter: En test til vurdering af semantisk hukommelse [Famous Faces: A Test for Assessment of Semantic Memory]. Copenhagen, Denmark: Hogrefe.
    Kertesz A. 1982. Western Aphasia Battery. San Antonio, Texas: Psychological Corporation.
    Leff A, Crewes H, Plant GT, et al. 2001. The functional anatomy of single-word reading in patients with hemianopic and pure alexia. Brain. 124:510–521. doi:10.1093/brain/124.3.510
    Leff A, Starrfelt R. 2014. Alexia: Diagnosis, Treatment and Theory. London, England: Springer Verlag.
    Leff AP, Spitsyna G, Plant GT, et al. 2006. Structural anatomy of pure and hemianopic alexia. J Neurol Neurosurg Psychiatry. 77:1004–1007. doi:10.1136/jnnp.2005.086983
    Lefly DL, Pennington BF. 2000. Reliability and validity of the Adult Reading History Questionnaire. J Learn Disabil. 33:286–296. doi:10.1177/002221940003300306
    Pelli DG, Waugh SJ, Martelli M, et al. 2016. A clinical test for visual crowding. F1000Res. 5:81. doi:10.12688/f1000research.7835.1
    Petersen A, Vangkilde S, Fabricius C, et al. 2016. Visual attention in posterior stroke and relations to alexia. Neuropsychologia. 92:79–89.
    Riddoch J, Humphreys GW. 1993. BORB: Birminghan Object Recognition Battery. Hove, United Kingdom: Taylor & Francis.
    Sand K, Robotham RJ, Martelli M, et al. 2018. Visual crowding in pure alexia and acquired prosopagnosia. Cogn Neuropsychol. 35:361–370. doi:10.1080/02643294.2018.1483325
    Shallice T. 2014. On compensatory strategies and computational models: the case of pure alexia. Cogn Neuropsychol. 31:1–15. doi:10.1080/02643294.2014.914023
    Starrfelt R, Habekost T, Gerlach C. 2010. Visual processing in pure alexia: a case study. Cortex. 46:242–255. doi:10.1016/j.cortex.2009.03.013
    Starrfelt R, Habekost T, Leff AP. 2009. Too little, too late: reduced visual span and speed characterize pure alexia. Cereb Cortex. 19:2880–2890. doi:10.1093/cercor/bhp059
    Starrfelt R, Shallice T. 2014. What’s in a name? The characterization of pure alexia. Cogn Neuropsychol. 31:367–377. doi:10.1080/02643294.2014.924226
    Warrington EK. 1984. Recognition Memory Test. Windsor, United Kingdom: NFER-Nelson.
    Warrington EK, Shallice T. 1980. Word-form dyslexia. Brain. 103:99–112. doi:10.1093/brain/103.1.99
    Wilson BA, Cockburn J, Halligan PW. 1987. Behavioral Inattention Test. London, England: Pearson Assessment.

    pure alexia; recovery; posterior cerebral artery stroke; alexia without agraphia

    Supplemental Digital Content

    Copyright © 2019 Wolters Kluwer Health, Inc. All rights reserved