The psychogenic theory of Bettelheim (1967) maintains that autism in young children is an emotional disorder resulting from the behavior of cold and unresponsive parents who harbor hostile impulses toward their child. As a result, the child develops autism as a defense against such parental behavior. Treatment involves separating the parents from their child, generally by placing the child in a warm and loving residential program in which he or she is given as much freedom as possible. Any indication that the child is receding into autism is countered with unconditional love and affection. Parents are allowed to participate in the treatment to help resolve the unconscious conflicts that are said to be typical of themselves as well as their child.

Developed by Niko Tinbergen, a 1973 Nobel Laureate in the field of ethology, and his wife, a long-time teacher of children with autism, this theory holds that autism is a result of inadequate bonding between the mother and her child (Tinbergen and Tinbergen 1972, 1983). Adherents to this view, which includes Welch (1988), a psychiatrist, also believe that the child with autism is overly fearful and frustrated because of the parents’ awkward and unsuccessful attempts at socialization, and that these defensive, emotional reactions over-generalize to social stimuli such as faces and facial expressions. They maintain that the child is too young and insecure to conquer his or her fears and frustrations due to the failure to adequately bond with the mother, thereby creating feelings of inadequacy and an accompanying sense of loss and loneliness. According to this theory, these problems can be resolved by the mother holding her child, even forcibly and for long periods of time, while repeatedly expressing words of love, devotion, and affection. This practice, they explain, aims to break down the child’s autistic symptoms.

Nearly 50 years ago, Rimland (1964) reviewed the evidence and arguments for Bettelheim’s psychogenic theory and concluded that it was an “inadequate and pernicious hypothesis.” Since then, adherents to the theory have all but disappeared, at least in the USA. Thankfully, no credible scientists blame autism on bad parenting any more. And, while bonding theory and “holding therapy” gained some popularity in the 1980s, they are also perspectives with few, if any, supporters today (Schreibman 2005). While they rightfully belong to a bygone era, we will see later on that these misguided theories persist as animus to discussions on the role parents play in the development of their child’s behavior, including autistic behavior.

Along with Kanner’s original observation and report of children with autism was his speculation that the course of the disorder was constitutional in nature (Kanner 1943). Later, Kanner (1949) emphasized that the constitutional deficits were chiefly in the social and emotional domains.

An extended version of Kanner’s position is given by Hobson (1989). Summarizing Hobson’s position, Matson (1994) wrote that, “…autism stems from constitutional limits on emotional reactivity that alter the necessary sharing of subjective interpersonal experiences. Such limits impede abstraction, symbolic representation of thought/feelings, and emphatic recognition of feeling and thought in other persons. Social and affective development is undermined, leading to secondary, lower-order deficits in cognitive and language functions” (p. 40). Although the Kanner–Hobson hypothesis—the ability to represent the thoughts and feelings of oneself and others is deficient in children with autism—is not well known per se, it is nonetheless the key element in the impaired meta-representation theory.

Popularly known as the “theory of mind” or “ToM” deficit, the impaired meta-representational viewpoint holds that normal social interaction, communication, imagining, pretending, and so forth have their origins in the growing capacity to represent the mental states of oneself and others (Wing 1989). A disturbance in the development of this capacity, according to Frith (1994), affects “the thinking of autistic people, making them unable to evaluate their own thoughts or to perceive clearly what is going on in someone else’s mind” (p. 117). For Malle (2002), the inability undermines the development of “all conscious and unconscious cognition” and leads to what Baron-Cohen (2001) calls “mind blindness.”

The impaired meta-representation theory is currently the most vigorous cognitive deficit theory of autism. Among the most active contributors are the British psychologists Frith, Baron-Cohen, Leslie, and Wing, who set for themselves the task of searching “…for a single cognitive component that would explain the deficits, yet still allow for the abilities that autistic people display in certain aspects of interpersonal actions” (Frith 1993, p. 110–111). They take the position that while autism is an organic disease that cannot be cured, much can be done to make life more hospitable for those who are afflicted.

According to Sternberg (1987), autism may be understood in terms of his “triarchic” theory of intelligence. The relevant subset of this theory pertains to the relationship between intelligence and the mental life of the child. For the child with autism, the knowledge acquisition process, which involves selective encoding (discrimination), selective combination (construction), and selective comparison (integration), is applied in a manner that is “misselective” with respect to the social environment. As Sternberg explains it, the child selects aspects of social stimuli in certain areas, especially the linguistic-symbolic area, in a manner that eventually produces symptoms of schizoid personality disorder, which includes impaired social behavior and interpersonal interaction as well as emotional deficits and cognitive limitations.

In a speculative effort to show how Piagetian theory might be applied to children with autism, Morgan (1986) suggests the child may have a permanent or at least a long-standing imbalance between accommodation, which includes the figurative function of presenting symbolic meaning, and assimilation, which includes the operational function of conceptualizing symbolic meaning (see Cowan 1978). Morgan further speculates that the child’s repetitive and restrictive motor behavior and lack of imaginative and symbolic play may be related to an arrest of certain operative functions at the sensorimotor level.

Morgan (1986) also believes that young children with autism may have an abnormal concept of object constancy. In his view, parents and other people are the least predictable “objects” in the environment. The child with autism resists people on this basis—unpredictability—and is instead drawn to, and forms attachments with, actual objects in the physical environment with predictable features, for instance, a toy car or a stuffed animal. Morgan asserts that an abnormality of this sort in object constancy undermines the formation of normal social attachments and interferes with the development of advanced cognitive activities, such as symbolic play, which he claims are necessary for the early development and elaboration of verbal and social behavior.

These three theories of childhood autism uniformly regard the observable behavior of a child as indicative of a deficiency or abnormality in a hypothetical mental process, an undetected neurological condition, or more likely, a combination of these constructs. In other words, the theories rest on unobserved and unverified events in both the mental and the neurological domains.

Consider the dominant cognitive deficit theory, the impaired meta-representation theory. Said to be localized in one of the least understood regions of the human brain, the orbitofrontal cortex (Baron-Cohen et al. 1994), ToM and the meta-representational processes it entails is said to enable a child “…to infer the full range of mental states (beliefs, desires, intentions, imagination, emotions, etc.) that cause action” (Baron-Cohn 2001, p. 174, italics added). This causal agent is revealed, and its workings evaluated, by a “false belief task” (Caruthers and Smith 1996) that is given to assess a child’s ability to predict the actions of another person on the basis of an “inferred mental state that differs from reality” (Bloom and German 2000). For some researchers, failing a false belief task is tantamount to a diagnosis of autism, and the child with autism, in turn, is described as having a deficiency in meta-representation, a defective ToM, a damaged orbitofrontal cortex, or a combination of these disturbances.

A less gratuitous interpretation of the false belief task is offered by Schlinger (2009), who writes that, “a certain level of verbal fluency is necessary to make an inference about what another person might be thinking, in addition to predicting what he or she might do” (p. 442). Studies have repeatedly confirmed this by showing a high, positive correlation between performance on a false belief task and the level of vocabulary development of the children tested for ToM (e.g., Astington and Jenkins 1999; Bretherton and Beeghly 1982; Happe 1995; Shatz et al. 1983). In one such study, by Happe, it was noted that, “Unlike normally developing 3- and 4-year-olds or nonautistically mentally handicapped individuals, subjects with autism in this sample needed to have a high level of verbal ability on the British Picture Vocabulary Scale (BPVS) in order to stand a chance of passing false belief tests. In addition, a very high level of verbal ability was found among those autistic subjects who passed the theory of mind tasks” (p. 853). Studies with deaf children and children with autism report similar rates of failure on the task (Peterson and Siegel 1999), and other studies comparing children with autism to children of typical development and children with intellectual disabilities show that failing the task is not limited to children with autism but instead is related to the sophistication of the child’s verbal behavior (Yirmiya and Schulman 1996). While “reading minds” and engaging in related “perspective-taking” behaviors are undoubtedly important to the acquisition of verbal and social behavior, and vice versa, it is difficult to see the value of casting the process by which this occurs in the light of meta-representational theory. Furthermore, the autism spectrum contains a very diverse array of deficits, which vary greatly from person to person, so to ascribe the cause of all of these deficits to a lack of perspective-taking ability (a specific skill repertoire) seems a bit overly ambitious.

The problem with the cognitive deficit theories is easy to spot. The basis for meta-representation theory, for example, is the accuracy of child’s verbal-vocal report of where another person will look for a hidden item that was moved without their knowledge on a false belief task. Transforming this direct observation of behavior into unobservable processes and hypothetical mechanisms is congenial to mentalism. As Moore (2003) describes it, theorizing with mentalism is a three-stage process involving (1) collecting observations of behavior (e.g., performance on a false belief task), (2) hypothesizing the process that appears to underlie the behavior (e.g., meta-representation), and then, after additional observations of behavior (e.g., more performances on false belief tasks), (3) constructing a formal theoretical account of the mechanism responsible for the behavior (e.g., ToM), including its neurological basis (e.g., the orbitofrontal cortex).

The theorizing process described above is common in cognitive psychology and Skinner criticized it throughout his long career as circular, mentalistic, reductionistic, and impractical (e.g., Skinner 1978, 1987; see also Reese 1996). To elaborate on these familiar criticisms would take us far afield (for a cogent summary, see Moore 2003). We will, however, develop a case against reductionism in the context of the behavior analysis theories of autism that give causal status to biological factors.

Both Sternberg’s (1987) and Morgan’s views are compatible with the impaired meta-representation theory and therefore are subject to the same criticisms. In addition, Sternberg takes the position that cognitive defects are antecedent to deviant social behavior. If one were to take a different position, such as that taken by Skinner (1957), that there is an overlapping relationship between cognitive, social, and verbal behaviors , then one would expect that deviant social behavior would be antecedent to impaired verbal and cognitive behavior.

The behavioral hypothesis by Ferster (1961) holds that an experimental analysis of the basic variables determining a child’s behavior will show how these variables operate to produce the particular kinds of deficits seen in a young child with autism. Such an analysis, Ferster claimed, would reveal not only the effects of the child’s autistic behavior on members of the family, and vice versa, but also how these behaviors and interactions maintain the deficits. He claimed further that the behavior of the parent can forestall or weaken the development of their child’s behavior through the reduction of reinforcement, extinction, noncontingent reinforcement, and, to a lesser extent, aversive control. This state of affairs could come about through the disruption of the parents’ repertoire (e.g., depression), the prepotency of other behavior (e.g., a strong desire to be active in community affairs), or escape from the child who has acquired aversive properties due to continual screaming, tantruming, and the like. Ferster concluded that, “All of the variables that might weaken the behavior of a child are directly or potentially observable. The data required are the actual parental and child performances and the specific effects on each other, rather than global statements such as dependencies, hostility, or socialization” (p. 455).