University of Edinburgh

Visual Impairment and Autism -What we know about causation and early identification

presented by Dr Hilary Cass
Consultant Paediatric Disability

Autism and Visual Impairment Conference, Edinburgh, 27 March 1996

1 Visual Impairment and Autism - A Cause for Concern?

Peter paid no attention to me or my voice and sat or lay on the picnic table absently mouthing a rubber toy. Occasionally he made an irrelevant statement. Once he sang, in perfect pitch, "Pussy cat, pussy cat where have you been?" There were no toys to which Peter had any attachment. When he showed transitory interest in objects, he brought them to his mouth, sucked on them and chewed them. He did not explore them with his hands, he did not manipulate them. I observed that when Peter lost an object he was mouthing, he showed no reaction to the loss and did not search for it. Selma Fraiberg, Insights from the Blind, (1977)

Following Kanner's description of childhood autism in 1943, there have been multiple descriptions of blind children presenting a clinical picture similar to autism in the sighted, with early reports citing stereotypes, autism or echolalia, and self-absorption as frequently occurring features. The above description from the writings of Selma Fraiberg, one of the pioneers in this field, will present a familiar picture to many of those working with children with visual impairment.

As early as 1964, Fraiberg and Freedman made the observation that "ego deviation" (as they termed this condition) was present in 25-30% of blind children. This of itself is a very alarming assertion which might have been expected to arouse considerable concern among parents, educators and doctors. Perhaps one reason that it did not give rise to widespread comment is the association between visual impairment and learning disability. It is, of course, hard to obtain accurate data on the frequency with which learning disability accompanies visual impairment, not least because visual impairment is greatly under-reported in children with severe and complex disabilities. However, given the relative rarity of conditions resulting in "uncomplicated" blindness, compared to the more frequent situation in which visual impairment is just one component of a more widespread disability (as can occur in prematurity, pre-natal infection, birth asphyxia, various syndromes, etc), one can give a conservative estimate that at least 50% of blind children might be expected to have learning disability. Looking at the relationship between learning disability and autism, Wing and Gould, in their large population-based study in Camberwell, found the triad of impairments which characterise autism in 56% of those with severe learning disability (IQ <50). It might therefore be surmised that since learning disability is a common accompaniment of visual impairment, and autism a common accompaniment of learning disability, the association between visual impairment and autism needs no further explanation or investigation.

2 Retrospective study: Developmental Vision Clinic, Wolfson Centre

In the Developmental Vision Clinic at the Wolfson Centre (Institute of Child Health), we decided to take this issue further by looking specifically at the developmental outcome of those children who did not have additional disabilities as part of their presentation, and who at first assessment appeared to be developing normally. We felt that if we focused on children whose visual impairment appeared to be an isolated abnormality and found a high incidence of autism even in this group, then we would be facing a problem which really did merit a high level of professional concern.

The study undertaken was retrospective and drew on records collected over 15 years in the Developmental Vision Clinic. Children attend this clinic for specific developmental advice by a team experienced in the assessment and management of children with severe visual impairment.

The relatively small number of children in the study group (figure 1) highlights that fact that children with documented uncomplicated visual impairment are rarer than those with more complex disabilities.

STUDY GROUP Total Of 615 Children seen.

102 Children fitted inclusion Criteria:  
Age < 16 months at first assessment Normal development at first assessment
Follow-up to at least age 21/2 No additional disabilities

Figure 1

The children in the study group were divided into three groups, based on the severity of their degree of visual impairment (table 1):

Figure 1


In order to monitor progress, children attending the clinic are routinely assessed using the Reynell-Zinkin Scales, a standardised tool developed specifically for use with children with visual impairment. For our study group we were particularly interested in documented progress on the three most intellectually loaded sub-scales; sensori-motor understanding (non-verbal cognition), verbal comprehension and expressive language. Figure 2 demonstrates the expected progress of a normally developing blind child, which should run just above the blind mean (since the scales were standardised on a population containing children with additional physical and developmental disabilities).

Normally Developing Blind Child

Figure 2

Figure 2

As well as reviewing progress on the Reynell-Zinkin scales, we also noted any concerns expressed by either the parents or assessor about the children's social development and/or behavioural status, as well as details of any social adversity factors.

Even after carefully confining our study group to those children who seemed to be of normal developmental potential, our findings were very much in line with those of Fraiberg and Freedman, and are documented in table 2:

table 2


31% of children who were blind throughout had a regression in their development occurring between the ages of 16-27 months, compared to only 5% of the group whose vision improved and none of the group with "better" (albeit still very poor) vision throughout. This represents a significantly greater risk of setback for the blind group compared to those with "better" vision throughout (p < 0.0001), and compared to those whose vision improved (p <0.05).

There were two broad patterns of developmental setback; Pattern A in which acquisition of language and cognitive skills ceased abruptly, and in some instances loss of skills occurred, and Pattern B in which skill acquisition continued but the children's social interactions and communication become increasingly disordered.

Pattern A is exemplified by Patient 2 (see figure 3), a little boy with Norrie's disease (an X- linked retinal dysplasia). By 10 months he was making pleasing progress, the only concern being that he had some eye poking. Between the ages of 1-2 years, he had a series of medical problems; he had multiple admissions to hospital with severe eczema, required surgery for an infected haematoma after a fall, and had two eye operations for raised intra-ocular pressure. Additionally he lost his residual vision. Despite these problems, by age 2 he was speaking in 5-6 word sentences and cognitive development was normal; however, he was noted to have very negative behaviours and frequent severe tantrums. Shortly after his second birthday he underwent a major regression; he became withdrawn, self-absorbed and lost all speech and comprehension of language. Constructive play ceased, and was replaced by screaming, eye poking and banging toys in an aimless fashion. He did not recover from this regression.

figure 3

Figure 3

Pattern B (see figure 4) is demonstrated in a girl with Leber's amaurosis, who by age two was well ahead of the blind norms for expressive language, the only minor problem being a degree of echolalia which was not considered to be unusual in a blind child. Between age 2-3, her expressive language plateaued (although it remained above the blind mean); however, with the benefit of hindsight it appears that her use of language (a feature which is not well assessed on the Reynell Zinkin Scales) was becoming increasingly disordered. Although she continued to make cognitive progress, by age 4 she was not initiating any meaningful verbal or non- verbal communication, and a speech therapist described her language as a "stream of consciousness".

figure 4

Figure 4 Common to both these groups were a number of features, as shown in figure 3.

It was notable that the parents or assessor frequently felt that the child's learning was being blocked by the "blindisms", rather than that the "blindisms" were a secondary phenomenon. FEATURES OF SETBACK

Emergence of worrying patterns of behaviour increasing self-absorption Loss of functional communication Echolalia "Blindisms" or other stereotypies which were felt to be blocking learning. Disordered patterns of behaviour usually pre-dated objective fall-off in scores on Reynell-Zinkin scales.

figure 5

Figure 5

3 The Thorny Question of Causation


Figure 6

In trying to understand the causation of any kind of developmental disability, it is necessary to consider both child-related factors and environmental factors. Additionally the dynamic issue of how these factors interact over time may have bearing on the problem.

One crucial child-related factor - the role of degree of visual impairment - has already been touched on. However, it is difficult to determine whether the increased risk associated with more severe degrees of visual impairment is mediated by the sensory deficit per se, or whether greater degrees of visual impairment are merely a hallmark of more extensive neurological involvement.

Reviewing the simplified diagram below, it will be observed that many different kinds of brain insult (group A) can result in visual impairment. Within this group of conditions (excluded from our study for the reasons discussed previously), the coexistence of autism might not be considered surprising. Conversely damage involving only the globe (group C) represents a relatively "pure" form of blindness, in that only the "mechanical" part of the visual apparatus is involved. Disorders which affect the retina and optic nerve fall between these two extremes, because although one might not anticipate brain involvement, there is now ample evidence that neuropathology is present in many cases. For example Dekaben found that 10 out of 13 patients with congenital retinal dystrophy had neurological complications such as epilepsy and cerebral palsy and Roberts-Harry et al found CT scans abnormalities in 22 out of 33 children with optic nerve hypoplasia. Children with anophthalmia and microphthalmia have a variable degree of involvement of the optic nerve, as well as the globe, and hence one would predict that those at the anophthalmic (and hence more severe) end of this spectrum are at greater risk of more widespread brain involvement.

Within our study group, all the affected children had a disorder involving the neural elements of the visual system (ie, the optic nerve and/or retina), whilst none of the children with a peripheral cause for their visual impairment (eg, buphthalmos, cataracts) suffered developmental setback. Furthermore, although 2 of the 5 children with anophthalmia were affected, all 11 children with microphthalmia had normal outcome. Initially this gives the impression that conditions which have greater risk of central nervous system involvement also lead to a higher chance of setback; however, because these are the very conditions which give rise to more severe visual impairment, it remains very difficult to divorce these two factors.

Turning to environmental factors, within our study group 60% of the affected children had some kind of social adversity factor (such as parental drug addiction, marital breakdown, severe illness, etc) compared to only 23% of the group with normal outcome (p < 0.05). However, this information needs to be evaluated with caution, since in a retrospective study it is far more likely that adversity factors would have been documented in children who were presenting a developmental cause for concern than in those who were doing well. Nonetheless, it is interesting to note that 4 of the 11 affected children made a partial (albeit far from complete) recovery with additional developmental and social support.

It was notable that all the children who suffered setback regressed between 15-27 months. This does seem to be a particularly vulnerable period of development, with a number of other conditions (eg, Rett syndrome, tuberous sclerosis, some cases of autism in the sighted) becoming apparent at around the same time.

This information can be integrated by the hypothesis that the onset of regression is triggered in neurologically vulnerable children by a constellation of environmental conditions (that are in some way suboptimal for a blind child) occurring at a critical stage in the development of attention control and social interaction (see figure 8).

Figure 7

Figure 8

4 A Worrying Dilemma

This explanation gives rise to a serious dilemma: On the one hand, many people working in the field of autism are understandably disturbed by any aetiological explanation which has an environmental component, fearing a resurgence of the outdated and damaging "refrigerator parent" theory. At the same time, many educators and therapists with extensive experience in the field of visual impairment are unhappy about rationalisations which are based on a pre-programmed neurological theory, feeling that this does not intuitively hold true for the children under their care.

Perhaps a partial answer to this problem lies in the question of whether the behavioural disturbance observed in visually impaired children is the same phenomenon as autism in the sighted, or whether we are observing a superficially similar end-point arising from two quite different processes. At this point it is necessary to take a step back and consider a fundamental question; namely, what is autism?

5 What is Autism?

Kanner's very astute account of autism was the first of a series of clinical descriptions which sought to define autism in terms of a minimum number of features which had to be present in order to make the diagnosis (see figure 9). The problem with these early descriptions was that they did not clearly define the basis or core features of the disorder; hence a quite non-specific or variable feature (such as presence of special skills or abnormal sensory perceptions) might rank just as highly in the necessary diagnostic criteria as the features which we now know to be far more crucial (such as poor joint attention behaviours and lack of communicative intent). The picture became clearer in 1979 when Wing and Gould advanced the concept of a core impairment in social interaction. This led on to the model of a closely related triad of impairments (in social interaction, social communication and play), with children falling on to a continuum in terms of their severity of impairment in these three areas. More recent tools used for the diagnosis of autism reflect (to a greater or lesser degree) the presence or absence of the core impairment, with the "gold standard" for accuracy of any rating scale being how well it identifies children who fit the definition for autism as outlined in DSM (Diagnostic and Statistical Manual) or ICD (International Classification of Diseases). However, it is important to remember that it is the presence of the core impairment which means that a child has autism, not the fact that he or she scores to a minimum level on a rating scale. This distinction becomes important when considering children with visual impairment, because although rating scales such as the Childhood Autism Rating Scale and the Autism Behaviour Checklist may have included visually impaired children in their standardisation samples, they were not designed or standardised specifically for use in this group.

Figure 9

6 A Dubious Inference

21 And Isaac said unto Jacob: 'come near, I pray thee, that I may feel
thee, my son, whether thou be my very son Esau or not.'

22 And Jacob went near unto Isaac his father; and he felt him, and
said: 'The voice is the voice of Jacob, but the hands are the hands
of Esau.'

23 And he discerned him not, because his hands were hairy, as his
brother Esau's hands; so he blessed him.
What happens when rating scales standardised on a largely sighted population are applied to children with visual impairment? Clearly some normally developing blind children will achieve the same scores as normal sighted children, a proportion of normally developing blind children will have "borderline" scores, and some blind children will score as sighted autistic children. This reflects the emphasis within these scales on: (a) Highly visually-dependent joint attention behaviours such as eye contact, referential eye gaze and pointing.

(b) Repetitive behaviours which may be normal in blind children (eg, rocking, eye poking.

(c) Language abnormalities which may be normal in blind children (eg, pronoun reversal).

(d) Orientating behaviours which may be normal in blind children (eg, smelling, touching, etc.)

The problem comes when a score in the abnormal range is then used to infer that a child has a core impairment in social empathy and therefore, by definition, autism (see figure 10). Of course, the child might well have such an impairment, but one cannot make this assumption without rating him or her on a scale designed to assess social development in blind children. Unfortunately at the present time such a scale does not exist, and indeed our knowledge of normal social development in blind children is woefully inadequate.

Figure 10

7 Social Development - Knowledge and Gaps

Social development is a complex phenomenon, made up of a number of different processes, as listed in figure 11: SOME OF THE PROCESSES THAT MAKE UP SOCIAL DEVELOPMENT Attachment behaviours Social communication Joint attention behaviours Peer interactions Social and moral conscience etc. Figure 11

It may be self-evident, but before any attempt can be made to study abnormal development, it is crucial that we have some normative data. This means that for each of the processes which constitute normal social development it is important to have an appreciation of the normal behavioural manifestations, the normal developmental sequence and the range of normality. Figure 12 outlines the steps which are necessary before a rational approach to early identification and remediation can be considered, with shaded areas indicating the major gaps in our knowledge.

Figure 12

The social development of the normal child has, of course, been the subject of a very extensive literature, with attachment being perhaps the best understood of the social development processes. To illustrate the paucity of our knowledge about normal development in visually impaired children, it is perhaps worth comparing our understanding of this one process in blind and sighted infants:

Table 3

Table 3 is based on the work of Selma Fraiberg, who documented in detail the development of a small sample of blind infants. It summarises just a few of the ways in which blind and sighted children manifest their attachment to parents. For example, within Fraiberg's sample the blind babies lacked the exclusive smile for parents which is so characteristic of normally developing sighted infants. However the did exhibit an "alternative" preferential behaviour which is not part of the repertoire of sighted babies, namely frequent manual exploration of parents' faces. Some less visually dependent behaviours (such as stranger avoidance) appeared at similar ages in both groups, whilst some highly visually dependent behaviours (such as reaching to mother) appeared at a later stage in the blind group. In summary therefore, the blind children could exhibit the same behaviours as the sighted children, either at the same time or later, or they could exhibit quite different behaviours.

Unfortunately available data on social development in blind babies (including that of Selma Fraiberg) is generally based on samples of around 6-12 infants, and hence the most that can be abstracted is a broad idea of behavioural manifestations and sequences, with little idea of the full range of normality.

Accepting that our knowledge of normal social development in blind babies is very sketchy, the next question to address is which social development processes can go wrong. It has taken a long time to reach an understanding that sighted autistic children do not have a global impairment in their social interactions, and this is a widespread source of confusion amongst both parents and professionals. In fact, very specific processes are affected, as outlined in table 4:

Table 4

One of the key areas of difficulty is in joint attention behaviours, and a recently developed screening tool, the CHAT, is showing promise as a method for detecting impairments of this kind at age 18 months. However, if the joint attention process is one of the most compromised areas of development in sighted autistic children, it is not difficult to appreciate how fragile this process might be in blind children. Early video material of blind children (Preisler, Moore & McConachie) highlights the fact that even normally developing blind children will have a very different experience of shared attention situations when compared with their sighted peers (see figure 13):


Parents may not recognise baby's subtle "cues" for joint attention (eg leaning towards object rather than pointing). Parents may talk about object that is not the current focus of the child's attention. Child will not have visual cues for timing entry into conversations. Isolation in pre-school settings because of difficulty sharing focus of other children's interest, unless specifically cued.

Figure 13

To sum up the current state of knowledge, we have a poor grasp of how normal social development proceeds in blind children, and given the blind child's very different experience of the world, it is in some ways more remarkable that the process frequently succeeds than that it sometimes fails. Whilst we might make the assumption that the autistic-like picture observed in a proportion of blind children arises from the same core deficit as that in sighted autistic children, we do not yet have enough data on the range of normal social development in blind children to be able to prove this assertion, not to identify early markers of incipient disaster.

8 Moving Forward: a Pilot study aimed at the development of a normative social communication scale for blind infants


Does s/he enjoy being cuddled? yes/no... if yes, it this:

How does s/he show this? If something interests her/him, does s/he try to share the experience with you? yes/no...if yes, if this:

Describe how s/he does this

Give an example
Figure 14

A clear priority in enabling progress in this field is the development of a normative scale for social and communicative development in blind children. At the Wolfson Centre the Vision Research Team is currently piloting a parental questionnaire aimed at identifying the normal developmental sequences for blind children without using sighted behaviours as a starting point. The questionnaire has been administered cross-sectionally to the pilot group, which consists of 20 sighted and 15 blind children aged up to three years, but we are aiming for longitudinal administration to a definitive study group. 34 questions have been selected to sample a range of social and communicative behaviours (see figure 14), as well as to elicit behaviours which may be abnormal or give cause for concern. All the questions have both closed and open-ended components (with standard probes) designed to elicit information about the incidence, frequency and mode of each behaviour. A major aim of the pilot study is to eliminate questions which are ambiguous or unhelpful in establishing a developmental sequence, so that the final scale will be somewhat shorter than the original.

A few more interviews are required before data collection for the pilot study will be complete, but preliminary analysis of the available data suggests that blind children are more delayed in their language acquisition and joint attention behaviours than their sighted peers (a finding which replicates previous studies 12, 14) and that even at comparable levels of language development, blind children have less modes available to them to support joint referencing. However, infants in the sample with even a small amount of residual vision exhibit social communication behaviours which more closely parallel those of sighted infants. If this finding is substantiated by further data is will give some basis for the apparently higher risk of autistic-like behaviour in the more severely visually impaired group described in the first part of this paper.

9 Conclusion

The field of visual impairment and autism is extremely confusing and this paper has perhaps raised far more questions than it has answered. Nonetheless, although the "tidy-minded" doctor likes to take a cause-effect approach to developing treatment approaches, this is often not the way that medicine advances, with progress in appropriate treatment and remediation often preceding our understanding of causation. For this reason it is quite possible to attack the problem from both ends and hope that somewhere in the midst of the confusion we are able to advance our management of this difficult condition to the advantage of the children in our care.

Both the retrospective and the pilot studies described in this paper were carried out in the Development Vision Clinic at the Wolfson Centre (Institute of Child Health) under the directorship of Dr Patricia Sonsken, Senior Lecturer in Developmental Paediatrics. Data and graphs from the retrospective study are reproduced with the permission of Archives of Disease in Childhood. The principal investigators on the pilot study are Naomi Dale (consultant clinical psychologist) and Kirsty Brough (psychology student), and thanks are due to both for their contribution to this paper.

10 References

Kanner, L (1943) Autistic disturbances of affective contact. Nervous child; 2:217-250.
Fraiberg, S (1977) Insights from the Blind. London: Souvenir Press
Fraiberg, S and Freedman, D (1964) Studies in the ego development of the congenitally blind. Psychoanalytic Study of the Child; 19:113-169.
Wing, L and Gould, J (1979) Severe impairments of social interaction and associated abnormalities in children: Epidemiology and classification. Journal of Autism and Developmental Disorders; 9:11-29.
Cass, H Sonsken, P and McConachie, H (1994) Developmental setback in severe visual impairment. Archives of Disease in Childhood; 70:192-196. BMA House, Tavistock square, London WC1H 9JR.
Reynell, J (1979) Manual for the Reynell-Zinkin Scales. London: National Foundation of Education Research.
Dekaben, AS (1972) Mental retardation and neurological involvement in patients with congenital retinal blindness. Develop Med Child Neurol; 14:436-444.
Roberts-Harry, J, Green, SH and Willshaw, HE (1992) Optic nerve hypoplasia: associations and management. Arch Dis Child; 65:103-6.
Schopler, E, Reichler, R, DeVillis, R and Daly, K (1980) Towards an objective classification of childhood autism: childhood Autism Rating Scale (CARS). Journal of Autism and Developmental Disorders; 10:91-103.
Krug, D, Arlick, J and Almond, P (1980) Behaviour checklist for identifying severely handicapped individuals with high levels of autistic behaviour. Journal of Child Psychology and Psychiatry; 21:221-229.
Baron-Cohen, S, Allen, J and Gillberg, C (1992) Can autism be detected at 18 months? The needle, the haystack and the CHAT. British Journal of Psychiatry; 161:839-843
Preisler, G (1991) Early patterns of interaction between blind infants and their sighted mothers. Child: Care, health and Development; 17:65-90.
Moore, V, McConachie, H (1994) Communications between blind and severely visually impaired children and their parents. British Journal of Developmental Psychology; 12:491-502.
McConachie, H, Moore, V (1994) Early expressive language of severely visually impaired children. Developmental Medicine and Child Neurology; 36:230-140.

11 Biographical Detail

Dr Hilary Cass is Consultant in Paediatric Disability at Harper House Children's Service, Horizon NHS Trust and the Hospitals for Sick Children, Great Ormond Street. After qualifying with Honours at Royal Free Hospital, London in 1982, she trained as a general paediatrician and then went on to develop her higher specialist training in the field of disability. This interest was initiated through a long-standing involvement with Kith and Kids, a London-based, self-help group for families of disabled children, and she remains as professional adviser and personal friend to many members of the group.

Apart from her role in investigating the causes of developmental regression in blind children, her other clinical and research interests include autistic regression in previously normal infants, cognitive and behavioural effects of early epileptic disorders, management of children with Rett syndrome and the efficacy of augmentative communication systems in children with cerebral palsy. She is also actively involved in the audit of service provision for children and adolescents with neuro-developmental disorders.

Additional responsibilities include her roe as BPA Representative on the Council for Disabled Children and advisory roles to a number of other voluntary bodies.