In a previous post – Autism, Genius, and Minicolumns – I wrote about a research paper by Dr Casanova et al comparing the brains of three prominent neuroscientists vs. controls. One of the findings in this paper was a similarity (and differences) between the minicolumnar structures of the neuroscientists and those with ASD. I speculated based on the results that one of the differences between the neuroscientists and those with ASD might have been related to spindle cells (VENs). Since then the monkeys have been busy typing, and have another post ready. This post will explore the potential links between VENs, the Anterior Cingulate Cortex (ACC), Anterior Insular Cortex (AI), the frontopolar cortex, and autism. I will suggest that together they may explain the developmental origins of ASD.
Von Economo Neuron Morphology and Connectivity
Von Economo neurons (VENs) are large, bipolar neurons located in layer V of Brodmann Area 24 of the anterior cingulate cortex (ACC), and anterior insular cortex (AI) (Watson et al, 2006). A key difference between VENs and pyramidal neurons is that the former have only a single large basal dendrite (vs. an array of smaller basal dendrites in pyramidal cells). VENs have been found in the ACC of humans and all of the great apes, and in the AI of humans and some other apes. They were not found in any other primate species, nor in 30 non-primate mammalian species examined by Nimchinsky et al (1999). VENs have also recently been reported in the brains of several marine mammal species (humpback whales, fin whales, killer whales and sperm whales) in both the ACC and the AI, plus – unlike humans and apes – in the frontopolar cortex, as well as being sparsely distributed in other areas.
VENs are significantly more numerous in humans than in other apes. In the human ACC they occur most often in clusters of three to six neurons, are located in layer Vb, and are conspicuous because of an otherwise low cellular density in this layer. They account for 5.6% of the number of pyramidal cells in layer V of the ACC (Nimchinsky et al, 1999).
Although their function has not been definitively established, VEN morphology and location offer definite clues. The VENs in the ACC and AI appear to be a single population of cells - i.e. undifferentiated by region – (Watson et al, 2006), potentially suggesting a common origin (see VEN Development below). Unlike pyramidal cells, with relatively sparse apical dendrites and highly branched basal dendrites, VENs have long, narrow radial arborization, with similar profiles in their apical and basal dendrites in terms of 'branchiness' and length (Watson et al, 2006). Cell morphology is significant in that neuronal shape is directly related to the computations performed by the cells. Both spines and branches of neurons can operate as computational compartments, and VENs have fewer of both compared with layer V pyramidal cells.
Narrow arborization also impacts VEN connectivity. Minicolumnar structure tends to result in input into a column being relayed rapidly in a vertical direction, but not on a horizontal dimension. The narrow VEN dendritic trees suggest that they usually receive neurotransmission only within their individual minicolumns. Watson et al (2006) suggests that VENs could be a specialization that facilitates rapid output of minicolumnar processing.
This role is also suggested by the size of VENs. Their somas (cell bodies) are on average 4.6 times larger than layer V pyramidal cells, suggesting that they possess large and rapidly conducting axons. From Allman et al (2001), "Because cell body size is probably related to the size of the axonal arborization, the axonal arborization of the spindle cells may be extensive and on a scale with encephalization. This observation suggests that the spindle cells may have widespread connections with other parts of the brain". Allman et al (2005) postulated that the function of the VENs may be to provide a rapid relay to other parts of the brain of a simple signal derived from information processed within the AI and ACC. While it has not yet been firmly established where VENs project, their connections with the rest of the brain may be substantial.
VENs appear relatively late in human development. They first appear in very small numbers in the 35th week of gestation. At birth only about 15% of the final postnatal population are present, with potentially 95% of the total VEN population being present by four years of age (see Fig 1). The source of VENs is still unknown, but they could arise either from differentiation of a pre-existing cell type, or via migration, potentially from the ventricles. In human infants they often appear in pairs and sometimes in vertical chains of three or four neurons, suggesting that they may be following an anatomical or chemical path. Sometimes in infants they have long, undulating leading and trailing processes that resemble flagella, which also suggests migration.
Fig. 1 – Source: Allman et al, 2005
Regardless of migration vs. differentiation, VEN emergence may be vulnerable to disruption during post-natal development, resulting in ‘dysfunctional’ consequences related to neuropsychiatric disorders (Allman et al, 2005). In great apes and humans, VENs are approximately 30% more numerous in the AI in the right hemisphere (rAI) than the left. But they are only about 6% more numerous in the rAI over the left in newborns, so this right hemisphere predominance emerges postnatally, and may be related to the right hemispheric specialization for social emotions. The fact that this 30% right preference is so consistent across post-natal humans and apes suggests that it is important for normal functioning, and that deviations could be dysfunctional. In MRI comparisons of left and right hemispheres in a large population of normal subjects, the cortical grey matter volume was also greater in the rAI, consistent with the rightward predominance of VENs in this area (Allman et al, 2005).
Further support for the possibility of disruption of VEN development can be extrapolated from the known potential for disruption of other postnatally generated neurons. Allman et al (2001) refers to the example of postnatally generated neurons in the dentate gyrus of the hippocampus as being vulnerable to many stress-related events, and the survival of these cells can be enhanced by enriched environments. As well, the survival of postnatally generated neurons in the olfactory bulb is dependent on olfactory stimulation and the presence of brain-derived neurotrophic factor (BDNF), which in rat pups is enhanced by maternal care. It is thus conceivable that environmental stimulation, stress, and quality of parental care could affect the survival and development of VENs during infancy, ultimately influencing adult confidence or dysfunction in emotional self-control and problem solving.
Anterior Cingulate Cortex
The role of VENs is probably best considered in relation to their location. The ACC is the frontal part of the cingulate cortex (so named because it resembles a ‘collar’ around the corpus callosum), and includes both the ventral and dorsal areas of the cingulate cortex. Dr Maclean, in his triune brain model, theorized that the mammalian brain evolved in a series of concentric layers. The underlying reptilian core is comprised of the brain stem and cerebellum. Next is the limbic system, associated with emotions, motivation, and the association of emotions with memory, followed by the neocortex in mammals. In this view the ACC was considered part of the brain’s reptilian core.
More recent studies have found that the mammalian brain is homologous with the dorsal part of the forebrain in non-mammalian vertebrates, and was thus a specialization that evolved from this area. This specialization involved the segregation of the cortex into layers with distinct inputs and outputs. The ACC is distinct from much of the neocortex, in that it lacks layer IV, which is one of several layers that receive input from the thalamus. But it has a well developed layer V, and is similar in this laminar specialization to the motor areas of the neocortex, which lie adjacent to it. This, combined with findings discussed below, suggest that the anterior cingulate cortex is a specialized area of the neocortex devoted to the regulation of emotional and cognitive behavior (Allman et al, 2001).
Devinsky et al (1995) described the ACC as the mechanism by which affect and intellect can be joined, stating that it can be viewed as "both an amplifier and a filter, interconnecting the emotional and cognitive components of the mind." It plays a crucial role in initiation, motivation, and goal directed behaviours, and is part of a larger matrix of structures engaged in similar functions, described as a rostral limbic system, including the amygdala, periaqueductal grey, ventral striatum, orbitofrontal and anterior insular cortices. The system formed by these interconnected structures assesses the motivational content of internal and external stimuli and regulates context-dependent behaviours (Devinsky et al, 1995). Another feature of the ACC is its diverse thalamic afferents and resulting ability to sample a wider range of thalamic input than any other cortical region, which is significant, given its contribution to response selection. The ACC is also connected with the prefrontal cortex and parietal cortex as well as the motor system and frontal eye fields, making it a central station for processing top-down and bottom-up stimuli and assigning control as appropriate to other areas of the brain.
The ACC can be divided into a ventral/rostral affect region and a dorsal/caudal cognition region. The affect division is involved in conditioned emotional learning, vocalizations associated with expressing internal states, assessment of motivational content, and assigning emotional importance to internal and external stimuli, and maternal-infant interactions. The cognitive division contributes to skeletomotor control and is involved in nociception and pain response. A key role is in response selection associated with skeletomotor activity and responses to noxious stimuli. While the cognitive area may be related to some affective processes, evidence suggests that its role in affect is secondary to its role in cognitive processes, such as response selection, that do not require affect (Devinsky et al, 1995). Evidence of the division can be seen from electrical stimulation of the ACC. Stimulation of the ventral part of the ACC produces intense fear or pleasure in conscious patients, while stimulation of the dorsal part of the ACC produces a sense of anticipation of movement, supporting a disassociation between avoidance behaviour and affect (Allman et al, 2001; Devinsky et al, 1995)
Fig. 2 – Centers of activation of the ACC from brain imaging studies. Note the overlap between areas activated by cognitive and emotional tasks. The cortex containing VENs is indicated by purple shading, with the gradient in shading indicating anterior to posterior VEN density. From Allman et al, 2002.
Evidence of the affective / cognitive split can be seen by fMRI during different Stroop tasks. Stroop tasks require subjects to respond in the presence of conflicting or confounding information. In a cognitive version, a person could be asked to report the number of words on a screen, with the words being numbers such as ‘three’ written four times. In the emotional version, the words are emotionally charged, such as ‘murder’ written four times. The cognitive stroop version activates the dorsal part of the ACC, while the emotional stroop activates the ventral part. Although the areas of activation in these types of tests are largely separated, there is an overlap, suggesting that the dorsal and ventral parts of the ACC are probably interconnected (Allman et al, 2001).
The affective region plays a role in a wide variety of autonomic functions, such as the regulation of blood pressure, heart rate, respiration, and visceral responses to stimulation can include nausea, vomiting, epigastric sensation, salivation, and bowel or bladder evacuation. It is also involved in vocalization associated with internal states (Devinsky et al,1995). Activation is also associated with the experience of intense emotion, such as anger, love, and lust, and states of intense behavioral drive such as pain, hunger, thirst and breathlessness are also linked to strong activity. It is the affective region that receives a strong projection from the amygdala (Devinsky et al, 1995), which probably relays negative, fear-related information (Allman et al, 2001). There is also evidence that discrimination of emotion in faces – which is related to social awareness - activates the affective region (Devinsky et al, 1995), as well as evidence of increased ACC functioning in individuals with higher levels of social awareness (Allman et al, 2002).
The dorsal (cognitive) part of the ACC is strongly activated during focused problem solving and the performance of cognitively demanding tasks, with activation increasing with increases in task difficulty, and the same area is activated during the experience of intense drives such as love or lust. The commonality is that both types of activity involve intense mental focus (Allman et al, 2002). Studies suggest that the ACC is involved in response selection when novel choices are required, rather than during practiced responses, and that more demanding tasks increase ACC activation, while less demanding tasks reduce activation (Devinsky et al, 1995). The ACC is engaged in early premotor events that require cognitive information processing and correct response selection where a movement may be needed, including the determination of whether movement is actually required. The common feature in these functions is that they occur well before movement occurs, and the ACC may play an essential role in the transition from premotor to behavioural states. Devinsky et al (1995) indicate that this is one of the primary contributions of the ACC to human behaviour.
The cognitive region of the ACC is also involved in the perception of pain. Frith and Frith (2003) indicate that first order representation of pain occurs in the caudal ACC, but that the subjective perception of pain is located in the rostral ACC. Devinsky et al (1995) suggest the ACC role in pain sensation may be a) to determine its emotional affect, b) to select a motor response (i.e. initiate behaviour to escape the pain) and c) to learn how to predict and avoid pain in the future – i.e. learning. Of note, ACC lesions in animal studies retarded acquisition of discriminative avoidance learning. Other evidence of an ACC role – along with other brain regions - in learning and memory is evidenced by the ACC's role (mentioned above) in response selection when novel choices are required, rather than during practiced responses.
Both AI and ACC activation also increase with the degree of uncertainty, and they may be involved in both adaptive decision-making and response and cognitive flexibility. There is also a link between ACC signal output and the recognition of errors, resulting in a change in ACC electrical output called ‘error-related negativity’. Studies involving monkeys have shown that ACC neuron activation is related to the expectation of reward, supporting the concept that the ACC is continuously monitoring changes in feedback from interactions with the environment that affect survival and reproduction, and initiating behavioral responses to maintain or improve these conditions. These findings indicate that the ACC's error recognition and correction functions predate the appearance of VENs (Allman et al, 2002).
The ACC also appears to play a role in social interactions. Structural or functional changes in the ACC can cause significant changes in social behaviour. Studies have noted changes such as blunted affect, apathy, impulsivity, disinhibition, aggressive behaviour with minimal or no provocation, psychosis, sexually deviant behaviour, disabling obsessions and compulsions, and impaired social judgment. (To be very clear, given that this post is ultimately about ASD, I am NOT suggesting that those with ASD manifest the above behaviours).
One case in particular, reported by Eslinger and Damasio (1985), followed a previously successful accountant who – following bilateral ACC and orbito-frontal damage after a tumour resection, was unable to maintain his job, marriage, and finances despite preserved ‘intelligence and memory’. His ability to interpret social cues and adapt to social situations were severely impaired. He could recall normal patterns of social behaviour in hypothetical situations, but was unable to execute correct actions in real life. Of interest, follow-up research (Damasio et al, 1990) determined that there was a disconnection between his intellectual understanding of affect and the autonomic expression of same (measured by skin conduction during exposure to emotionally charged visual images). Devinsky et al (1995) speculated that the lack of physiological emotional cues may contribute to the patient’s ability to appreciate the emotional significance of stimuli. Of note too, area 24 lesions in infant monkeys impairs the separation cry when the infant is removed from the mother, and disrupts the mother’s ability to attend to the infant. (see Anterior Insular Cortex below for a discussion of interoception).
The ACC may also have a role in speech. ACC blood flow is elevated during the processing of words, suggesting a role in word and sentence selection. Devinsky et al (1995) suggest this is part of a broader response selection function rather than a specific language function or a linking of words to emotion, as this processing occurs in the dorsal (cognitive) division of the ACC. Other evidence of an ACC role in speech comes from akinetic mutism, which is most often associated with bilateral ACC lesions. One patient reported by Damasio and Van Hoesen (1983), who recovered from a left ACC stroke, initially had no spontaneous speech but was able to repeat. "She commented that ‘she did not talk because she had nothing to say. Her mind was ‘empty’. ‘Nothing mattered.’" During her initial illness she could follow the doctors’ conversations but ‘felt no will to reply’ to questions. This was interpreted by the authors as indicating that such lesions cause a ‘profound behavioural disturbance', preventing the normal expression and experience of affect. As mentioned by Allman et al (2001), patients with lesions in the ACC show reduced levels of spontaneous behaviour and willingness to act. Another recent paper (Chiung-Chih Chang et al, 2007) reported two cases of right ACC stroke that resulted in speech initiation problems. The authors suggest that in some cases, speech initiation requires the participation of the right ACC in addition to the language network in the left hemisphere. This too is probably more related to response selection than language function.
Overall, the ACC appears to play a crucial role in the brain, linking emotion, cognition, and response selection, and coordinating these capabilities with the rest of the brain. VENs may enhance the ability of the ACC to relay information to other parts of the brain. Layer V neurons typically relay the output of cortical processing to other cortical areas and subcortical structures, and the ACC has particularly large layer V pyramidal neurons (Devinsky et al, 1995). VENs may enhance the projection capabilities of the ACC, further enabling it to connect to the larger human brain. Allman et al (2001) indicated that VEN size across humans and great apes varies as a function of relative brain size - unlike pyramidal neurons in layer V and fusiform cells in layer VI of the ACC. VEN size and number are therefore probably an evolutionary response to the requirements of encephalization, responding to both increasing distance and increasing connectivity requirements.
ACC and the Development of Executive Awareness and Function
The ACC has also been linked to executive processes. Executive function is traditionally an umbrella term covering functions such as planning, working memory, impulse control, inhibition and mental flexibility, as well as the initiation and monitoring of action (Hill, 2004a). Frith and Frith (2003) indicate that many kinds of executive tasks are known to activate the ACC. They refer to a meta-analysis of executive tasks conducted by Duncan & Owen (2000), placing the peak activation of most of these tasks in the rostral cingulate zone (the same area activated by Stroop-like tasks). While Frith and Frith (2003) indicate that executive function is located separately in the brain from the mentalizing or theory of mind region, Hill (2004b) notes that studies indicate that executive function predicts performance on ToM tests, but not vise versa, suggesting that there is a complex relationship between the two (discussed further below).
It has been proposed by Posner and Rothbart (1998), that the ACC is involved in the maturation of self-control and executive function as the individual progresses from infancy to childhood to adulthood. The ACC is involved in evaluating pain, from a distress rather than a sensory perspective. Posner and Rothbart indicate that negative affect may be controlled by attention (e.g. distraction, as recognized by anyone trying to comfort an infant by attempting to focus their attention on other stimuli), suggesting that the ACC’s direction of attention develops as a means of coping with or controlling feelings of distress. They state that "amygdala-cingulate interaction might be a reasonable candidate for the earliest form of self-regulation in the infant". Initially this control of orienting is at least partially externally (e.g. care-giver) driven, but over time more direct control of attention shifts from care-givers to infants themselves, and infants become more involved in attempting to solicit adult attention.
Posner and Rothbart go on to state that "It seems likely that the same mechanisms used to cope with self-regulation of emotion are then transferred to issues of control of cognition during later infancy and childhood." From executive attention follows the early development of behavioural and emotional control. The authors demonstrate a developmental flow – supported by experimental results - from visual conflict resolution related to eye position (e.g. a transition from reaching requiring line of sight) to the generalization of this capability. Conflict resolution may further develop into error monitoring and voluntary inhibitory control over behaviour - including error correction - commensurate with ACC development.
As Posner and Rothbart (1998) wrote:
"The effort to develop ways of controlling distress provides a locus of control in the cingulate which may, step-by-step, generalize to other situations where conflicting demands must be resolved. Many years are devoted to development of systems of self-regulation. Indeed it seems likely that this development continues into adolescence and may be open to change in adult life."
This hypothesis is supported by the steady increase in the metabolic activity of the ACC from childhood to young adulthood, as well as studies linking ACC size in children and the ability to perform tasks requiring focal attention control.
Anterior Insular Cortex
The anterior insular cortex – the other area in the brain containing VENs – is known, along with the ACC, to have an important role in interoception, i.e. the subjective awareness of inner feelings. As Bud Craig (Craig, 2004) wrote, "The concept of ‘interoception’ was classically restricted to visceral sensations, but recent neuroanatomical and neurophysiological results indicate that sensations related to the ongoing physiological condition of all organs of the body – muscles, joints, teeth, and skin as well as the viscera – are processed together." These inputs are ultimately relayed from the thalamus to the insular cortex, and give rise to feelings such as pain, temperature, itch, muscle burn, visceral sensations, hunger, thirst, taste, and even sensual touch. Craig indicates that these feelings represent ‘the material me’, and the somatic-marker hypothesis of consciousness (proposed by A.R. Damasio) suggests that the sensory representation of the body is the basis for the mental representation of the sentient self, allowing the brain to distinguish the inner world from the outer world. In this theory, individual differences in emotional awareness are directly related to differences in the capacity for interoceptive feelings.
Research has demonstrated that right anterior insular cortex (rAI) activation and size are uniquely correlated with the subjective awareness of internal feelings of human beings. Sensory representation of the physiological condition of the body is initially on a same-side basis, and is then remapped to the rAI through the corpus callosum, supporting a role of the rAI in subjective feelings and the awareness of the physical self as a feeling entity, while thalamic inputs to the ACC produce behavioural drive (Singer et al, 2004). The rAI and rACC also appear to play a role in the integration of information about oneself linked to the process of visual self-recognition (Devue et al, 2007). Given rAI activation (selectively or in conjunction with the ACC) during many emotions, including anger, happiness, sadness, disgust, and lust, as well as by music, this supports the somatic-marker hypothesis and the link between interoception and "the emotional feelings that characterize human sentience" (Craig, 2004). The implication is that individual differences in subjective interoceptive awareness, and by extension emotional depth and complexity, might be expressed in part by the degree of expansion of the rAI.
Further, the rAI appears to be co-activated with the ACC when subjects experience social emotions (which as Craig (2004) pointed out, can be linked to physiological changes) such as guilt, embarrassment, violation of social norms, distress resulting from social exclusion, and humour, as well as when engaging in deception (Watson et al, 2006; Singer et al, 2004). Singer indicates that this may involve a simultaneous generation of both a feeling and an emotional motivation, along with its associated autonomic effects. I would suggest that this would be self-reinforcing, as the autonomic effects would feed back into interoceptive awareness. Both areas are activated when subjects view an image of a loved one compared with an acquaintance, suggesting that these structures may be involved in bonding (Allman et al, 2005). Further evidence for a role in bonding is provided by the presence of vasopressin 1a receptors in the VENs in these areas (see below). Further, Singer et al (2004) also found that the AI (bilateral but with peak activation in the rAI) and ACC (along with the brainstem and cerebellum) were activated not only when a subject receives pain, but also by a signal that a loved one experiences pain.
This indicates first a differentiation within the neural systems between the actual sensory experience of pain and its subjective experience and interpretation. But further, given the link of the ACC and rAI to the experience of a wide range of emotions, it also supports their role – independent of internal and external sensory inputs - as the neural basis of understanding both our own feelings and the feelings of others. Further evidence of the role of the AI and ACC in interpreting the emotion of others comes from Wicker et al (2003). In an fMRI study they found that observing the emotional facial expression of disgust in others activated the same sites in the AI and to a lesser extent in the ACC as activated by one’s own experience of disgust. Thus the authors suggested that observing an emotion in others activates the same neural representation of that emotion in us. Singer et al (2004) suggests that the ability to understand the feelings of others and to empathize evolved out of our own ability to subjectively represent our internal body states. Further, this capability is a necessary part of our ability to mentalize, i.e. to understand the thoughts, beliefs and intentions of others.
Given the existence of the capability for self-understanding, it is logical (in hindsight) that any evolved ability to understand others would be based on utilizing the same existing capability used to understand ourselves, rather than creating the functionality anew elsewhere in the brain.
The frontopolar cortex is another area of specialization in hominoids, especially humans. Brodmann Area 10 is large and well developed in humans, less so in great apes, and is much smaller in gibbons and monkeys. Within hominoids it declines in size in the same order as the decline in density of VENs (Allman et al, 2002). The lateral part of area 10 appears to be involved in episodic (related to specific events) as opposed to semantic (general knowledge) memory, and part of area 10 below this area is activated in the choice between smaller but more probable rewards vs. larger but more uncertain rewards, as is the ACC (i.e. response selection). The medial part of area 10 is activated when subjects are presented with emotionally charged moral dilemmas requiring choices that affect the lives of others. The medial and anterior areas are activated when subjects develop a successful decision-making strategy related to rewards. This activation may be related to the recollection of the outcome of recent attempts (episodic memory), the assessment of reward probability, and the choice of a strategy for the next attempt (Allman et al, 2002).
Strange et al (2001) indicates that the frontopolar cortex is activated during complex cognitive tasks, especially (along with multiple other regions) reasoning tasks, as demonstrated during the Wisconsin Card Sorting Test (WCST), the Tower of London task, inductive and probabilistic reasoning tasks, probabilistic classification, and the Raven’s progressive matrices test. There is evidence that it is involved in intentional or explicit rule induction - including generating and testing hypotheses about relationships between stimuli - when abstract structures or rules are needed, but that once a rule is learned then other areas are involved in its application. Damage to the frontopolar cortex can produce greater WCST sorting category switching impairments than damage to the dorsolateral prefrontal cortex (DLPFC), another key WCST-activated area, with frontopolar cortex activation and dysfunction linked to rule changes and learning and DLPFC activation and dysfunction linked to rule application. This is in line with another study cited by Strange et al (White and Wise, 1999) that the dorsal, ventral, and dorsolateral prefrontal cortices are involved in guiding behaviour according to previously learned rules. The frontopolar cortex may also mediate switching between different executive processes (Strange et al, 2001) .
The frontopolar cortex is also involved during working memory tasks, including when working memory requirements approach and exceed people’s short term memory limits, or when memory is split between tasks, as well as in ‘branching’ (a process in which an overall goal must be maintained while working on sub-goals), and the evaluation of retrieved episodic memories (Strange et al, 2001). The authors suggest that frontopolar cortex activation occurs in high level tasks that involve planning and executive control of cognitive functions, in particular those tasks that require a strategy or evaluative process to be applied to information held in short term or working memory, for example to test hypotheses on multiple items during rule induction. Decety and Jackson (2004) also suggested a potential frontopolar cortex role in the evaluation of self-generated responses, including when a task requires monitoring and manipulation of internally generated information.
Area 10 may also have a wider role along with the ACC in executive function and the behavioural maturation of self-control (discussed above). Strange et al (2001) saw a wider frontopolar cortex role in switching between different executive processes, a role which Posner and Rothbart indicates is also linked to the ACC. Both areas are activated when engaging in tasks requiring episodic memory – specifically in post-retrieval analysis (Strange et al, 2001) - and Allman et al (2002) suggest that an important part of the process of developing self-control and behavioural maturation is the ability to use past experience as a guide in responding to current events. Given ACC activation in these tasks, Allman et al (2002) suggest a functional linkage between the two areas, in which the ACC monitors the current state of reward and punishment and signals the need for behavioural adaptation, while area 10 compares the current state with past experience and - linking this to the consideration of the well-being of others (as per activation during emotionally charge moral dilemmas affecting others) - makes choices governing future behaviour. Based on Strange et al (2001)’s findings and White and Wise (1999)’s interpretation of the role of the prefrontal cortex, it may be possible to expand this to include a frontopolar cortex role in learning, with this knowledge then being passed on to the rest of the prefrontal cortex for continued execution.
Area 10 may also have a role in joint attention. Williams et al (2005) developed a video stimuli that, when watched by a subject, stimulated an experience of joint attention that could be monitored in an fMRI environment. The result was activation of the ventromedial frontal cortex, the left superior frontal gyrus (area 10), cingulate cortex, and caudate nuclei. The authors indicated that the VMFC has consistently shown activation during tasks involving the attribution of mental state. They further suggest that area 10 may serve a cognitive integration function, which in joint attention seems to utilize a perception-action matching process. The authors see the links between the areas above as evidence that the areas engaged in joint attention also serve a mentalizing function.
Allman et al (2002) indicates that the size of the dendritic arborization and the number of synapses of pyramidal neurons in area 10 is greater than in any other cortical area, suggesting an integrative role of this area fitting with that suggested by Williams et al (2005) above, and also by Strange et al (2001). This enhanced connectivity would presumably increase the impact of the ACC’s links to area 10 through subsequent activation of other areas. Interestingly enough, the finding of VENs in the frontopolar cortex of certain large marine mammals (see above) may also support Allman et al (2002) hypothesis of an evolutionary linkage between area 10 and the ACC, with potentially the same chemical signaling of VEN proliferation in the human brain being extended in the brains of the marine mammals in question.
The Role of VEN Neurotransmitter Receptors
The role of VENs may also be deduced from their neurotransmitter receptors (see Allman et al, 2005), which suggest that they are involved in the formation of social bonds and the anticipation of reward and punishment in uncertain conditions. They are part of a limited set of AI and ACC neurons in layer V that possess the vasopressin 1a receptor, which in rodent studies is strongly linked to the formation of social bonds. They are rich in dopamine D3 receptors, a high affinity dopamine receptor that has been proposed to signal the expectation of reward under uncertain conditions. AI and ACC activation increases with the degree of uncertainty, and the suggestion is that these areas are involved in adaptive decision-making and cognitive flexibility.
The serotonin 2b receptor is also strongly expressed in VENs. While it is rarely expressed elsewhere in the central nervous system, this receptor is also strongly expressed in the human stomach and intestines, where it promotes contractions of the smooth muscles responsible for peristalsis. Serotonin might serve as an antagonistic signal to dopamine, with serotonin signaling punishment and dopamine signaling reward. Activation of the serotonin 2b receptor on VENs might be related to the capacity of the activity in the GI tract to signal impending danger or punishment (literally ‘gut feelings’) and thus might be an opponent to the dopamine D3 signal of reward expectation. The result of these processes could be an evaluation by VENs of the relative likelihood of punishment vs. reward and a role in ‘gut level’ or intuitive decision-making in a given behavioral context.
Given the ACC and AI role in interoception discussed above, the presence of serotonin 2b receptors on the VENs that are otherwise rare in the brain, but common in the viscera, suggests an extension of the concept that the ACC and AI are monitoring activity in the gut. Allman et al (2005), suggests that these receptors might represent a transposition of this function from the gut into the brain, which would enable the organism to react more quickly to threatening circumstances than if it depended solely on monitoring sensations arising from the gut.
VENs and Intuition
Allman et al (2005) proposed a role for VENs in intuition. They wrote that:
"Intuition is a form of cognition in which many variables are rapidly evaluated to yield a fast decision. Typically we are unaware of the logical steps or assumptions underlying the process although intuition is based on experience-based probabilistic models. We experience the intuitive process at a visceral level. Intuitive decisionmaking enables us to react quickly in situations that involve a high degree of uncertainty which commonly involve social interactions. Frequently we do not have the luxury of sufficient time to perform deliberative cost-benefit analyses to determine the most appropriate course of action, but instead must rely on rapid intuitive judgments."
The ACC and AI are both active when individuals make decisions under a high degree of uncertainty. They are also involved in the subjective experience of pain, which is magnified by uncertainty, plus when subjects experience guilt, embarrassment, and engage in deception. They are also active in humour, trust, empathy, and the discrimination of the mental states of others. All of these social emotions are influenced by the degree of uncertainty involved. Allman et al (2005) hypothesize that "VENs and related structures integrate the probability of reward and punishment derived from many inputs and enable individuals to make quick, intuitive decisions that enable them to adapt to rapidly changing conditions". They would also be involved in relaying this output to other brain structures. As such, VENs would be an adaptation supporting the increased complexity of hominid - and especially human - social networks.
Minicolumns – Setting the Stage
Research by Casanova et al has indicated that those with ASD have a higher number of minicolumns than average, but that those minicolumns are of a narrower than average width, with smaller neurons but the same average number of neurons per minicolumn. The net result is a brain structure that skews in favour of processing stimuli that require discrimination, potentially at the expense of generalizing the salience of a particular stimulus. Smaller and more densely packed minicolumns could also allow for more complex information processing.
These attributes come at a potential cost. The reduction in width is a result of a reduction in the minicolumn’s peripheral zone of inhibitory and disinhibitory activity. The inhibitory fibers act to keep stimuli within individual minicolumns, and the reduction in this space increases the chance of stimuli overflowing to adjacent minicolumns, providing an amplifier effect and potential hypersensitivity. Narrower minicolumns may also result in an increased number of minicolumns per macrocolumn, which can also result in an amplification of thalamic input, and as each minicolumn’s response to thalamic input is modulated by the activity of neighbouring columns, a reduction in GABAergic inhibitory activity could also result in a loss of inhibition and greater amplification. Stimuli ‘spill’ and greater amplification could result in the increased incidence of seizures in autistics.
An additional factor is the reduction in neuron size, which reduces the ability of neurons to sustain connections over distances. Smaller neurons result in a metabolic bias favouring shorter connections at the expense of both longer distance and inter-hemispheral connectivity. The result is that autistic brains have a bias towards local (intra-regional) over global (inter-regional) connectivity and processing. Short intra-regional processing functions include mathematical calculations and visual processing. Cognitive functions that require inter-regional processing would be less metabolically efficient, including language, face recognition, and joint attention (Casanova - Abnormalities Of Cortical Circuitry In The Brains Of Autistic Individuals). Given the high metabolic cost of the brain (2.5% of our body weight but 22% of our resting metabolism - Leonard and Robertson 1992, p 186), smaller neurons may be a response to resource constraints.
Of note, while reduced minicolumnar width appears to be a prerequisite for ASD, the reported minicolumnar widths found within autistic brains are still within the normal distribution of minicolumnar width, albeit at the tail end (Casanova, 2006). In other words, people with narrow minicolumnar widths – i.e. a pre-autistic brain - are not necessarily on the ASD spectrum, but instead, minicolumnar width is part of normal human variation (neurodiversity). Something further is required to cause ASD.
VENs and ASD
Given the roles of the ACC, AI, and VENs discussed above, any disruptions in VENs in the autistic brain could potentially be expected to have an impact on the functions performed by the ACC and AI. And disruptions appear to exist.
Allman et al (2005) noted a study demonstrating the portion of the ACC in the right hemisphere containing VENs was reduced in volume in individuals with ASD compared with matched controls. A diffusion tensor imaging (DTI) study found that long distance fibre connections in the white matter adjacent to the ACC, including the area which in normal subjects carries the axons of VENs, was disordered in ASD subjects. Postmortem analysis of ASD brains by Kemper and Bauman (1993) found that the ACC was poorly laminated. Another study found isolated pockets of neurons in the white matter of the frontal cortex of ASD subjects, suggesting a defect in neuronal migration (which might also impact VENs). Further, Allman et al (2005) makes reference to unpublished data (also Allman et al) from autopsies of a 9 year old boy and 9 year old girl with ASD which found heavy concentrations of VENs in the white matter and extending through layer VI into layer V in the AI. VENs in these subjects were also located medially to their normal AI location.
Functional imaging data also suggests a connection between the AI and ASD. A study asking high functioning autistics and controls to identify the mental state of photographed individuals found rAI activation in controls but not in those with ASD. Allman et al (2005) also noted reports of reductions in measurements of embarrassment and empathy (both of which activate the AI and ACC) in autistic subjects, although they also noted that there is a considerable overlap between autistic and normal subjects.
Allman et al (2005) hypothesizes that the social disabilities in ASD are partially due to abnormal VEN development. The authors state that:
"our hypothesis is that the VENs and related structures integrate the probability of reward and punishment derived from many inputs and enable individuals to make quick, intuitive decisions that enable them to adapt to rapidly changing conditions. Because social emotions by their very nature involve considerable uncertainty, and because social interactions are often of a rapidly changing nature, an impairment of the VEN system would be predicted to compromise social functioning. The lack of quick social intuitions is a key deficit in autism spectrum disorders."
Allman et al (2005) also link the possibility of a reduction in ability to quickly process uncertainty with the desire for sameness and routine in ASD. "In short, we believe that the VEN system for rapid, intuitive responses in situations involving considerable uncertainty is impaired in autism spectrum disorders." They are clear that they are not proposing that VENs are related to Theory of Mind, but rather, that VENs might serve as input into the neural system that creates mental models of the thinking of others.
A recent paper by Kennedy et al (2007) also explored VENs in ASD, originally hypothesizing a reduction in the number of VENs in the ASD brain. They sampled and extrapolated the number of VENs in 4 ASD brains (three lAI and one rAI) vs. five controls (two lAI, one rAI and two with both hemispheres). But rather than finding a reduction in VENs in ASD brains, they found similar overall numbers, once the ASD outlier (lAI) was omitted. While the low sample sizes render any comparisons anecdotal, I did note that the VEN counts in the ASD 3 year old (lAI) were quite high compared with most of the controls, and the ASD outlier count was 27% higher than the next highest (control) count and 63% higher than the average adult control count (three left and two right hemispheres). I also noted that the VEN numbers found (even considering most counts are for a single hemisphere) are substantially lower than the equivalent AI counts for both hemispheres indicated by Allman et al (2005) (see Figure 1 above), despite the fact that both sets of numbers were generated through stereological counts. Even so, the common methodology used for all samples in the Kennedy et al (2007) paper could allow for a valid comparison between samples. The authors also noted that they observed no qualitative difference in the size of VENs in control and ASD samples at either the younger or older ages.
My takeaway from the paper, regarding ASD and the AI, is that there may be issues regarding VEN quantity, migration, and ‘pruning’ that require further investigation. I would also note, as did the authors, that their findings are not necessarily applicable to the ACC.
Going Further – Implicating VENs, ACC, AI, Frontopolar Cortex and ASD
When approaching the subject of ASD, one of the benefits of being a parent rather than a neuroscientist is one can speculate without risk of damaging one’s professional reputation. So here goes. I would suggest that there is a link between VENs, the ACC, AI, frontopolar cortex, the narrower minicolumn hypothesis, potentially the cerebellum, and ASD.
As mentioned above, the major period of VEN proliferation is during the post-natal period, and these cells are found in the ACC and AI. I would speculate that some of the major deficits in ASD may result from disruption of VEN development and/or development of the ACC and AI. VEN development issues may precede or follow ACC/AI disruptions, but evidence (e.g. as per Allman et al, 2005) suggests that VEN development may be impacted, and the results may be significant.
VENs may be the class of neurons that compensate for the bias inherent in smaller pyramidal neurons in the reduced minicolumnar width pre-autistic brain. Because of their size and projection capabilities, VENs may enable the ACC and AI to adequately connect to and influence the development of the brain despite the bias of smaller pyramidal neurons against more global connectivity (as per the minicolumn hypothesis). Belmonte and Carpenter (1998) noted findings by Bauman and Kemper (1985) of small, closely packed cells in the ACC of autistic brains, presumably including the pyramidal neurons in layer V. I would suggest that this is in keeping with Dr Casanova’s minicolumn findings discussed above. A functional population of VENs could presumably compensate for the bias in favour of shorter connections in these smaller cells. But disruption in the network of VEN cells would reduce this compensatory effect.
To be clear, I’m not suggesting an ‘all or nothing’ impact from variations in VENs in ASD. Instead, I would suggest that variations in VEN location and numbers could result in variations in connectivity and development. Three thoughts come to mind. First, a higher or lower number of VENs could result in ‘over-connectivity’ or ‘under-connectivity’, but dysfunctional connectivity nonetheless. As per Allman et al (2005):
"This right hemisphere VEN predominance [in the AI] may be related to the right hemispheric specialization for the social emotions. The fact that this 30% right preference is so tightly regulated and consistent across postnatal humans and apes suggests that it is important for normal functioning and that deviations from this ratio could be dysfunctional."
In addition, as mentioned in Allman et al (2002), neuron death is also necessary for normal development, and Allman et al speculated that excessive VEN survival could contribute to psychiatric disorders associated with over-activity of the ACC, such as OCD, which is characterized by excessively active vigilance and error-correcting behaviour. Higher VEN counts in ASD might also have an impact. In contrast, VENs are noted to be particularly vulnerable to degeneration in Alzheimer’s disease, with a loss of 60% of VEN neurons noted by Nimchinsky et al (1999).
Second, altered migration could result in dysfunctional connectivity. As mentioned above, Allman et al found heavy concentrations of VENs in the white matter and extending through layer VI into layer V in the AI, and noted that VENs were located medially to their normal location. Given the expectation that neurons will connect and function or perish, mislocated pockets of VENs would presumably still be functional, leading to the possibility of arbitrary and errant connectivity, with potentially disruptive consequences.
Third, variations in ASD may result from different ‘autisms’ – i.e. different etiologies and vulnerabilities mixing with variations in environmental factors (‘environment’ being defined broadly) - and result in variations in presentation. For example (this is pure speculation), higher levels of BDNF in ASD may signal an increase in VEN birth and/or survival, at the expense of necessary ‘pruning’. Various autism etiologies may lead to variations in VEN proliferation, brain connectivity, and – ultimately – ASD presentation.
The cerebellum may also play a role in ACC and AI issues. Brain imaging studies have shown that the cerebellum plays an important role in selective attention (i.e. the focusing on a specific aspect of a scene while ignoring other aspects - Wikipedia). Belmonte and Carper (1998) indicate that purkinje cells of the cerebellar cortex are the sole inhibitory input to the deep cerebellar nuclei. Loss of appropriate inhibition could produce activation of the deep nuclei and result in nonspecific activation of the cerebral cortex, resulting in impairment of attentional functions. Purkinje cell counts are known to be reduced in those with ASD. In contrast, Bischoff-Grethe et al (2002) suggests that the cerebellum is involved in response reassignment downstream, based in part on input from the ACC. Ravizza and Ivry (2001) also suggest that attentional deficits regarded as being due to cerebellar dysfunction may be at least in part due to issues related to coordinating successive responses.
Regardless of the cerebellum’s position in the attentional loop, any impairment in cerebellar function relating to attention shifting – input and/or output related - would presumably negatively affect the ‘success’ of the ACC in ultimately directing attention, and could result in significant developmental consequences. From a VEN perspective, the developmental and/or migratory stimuli and pathways are as yet unknown, but one could reasonably suggest that alterations in connectivity between the cerebellum, AI and ACC might potentially affect the chemical messaging required for VEN development.
Implicit (and sometimes explicit) in the discussion below are four assumptions. The first assumption, which is presumably fairly widely held, is that cognitive evolution has historically been driven by tests of social fitness. Big brains are metabolically very expensive, and intelligence – presumably inherent in a larger brain - is a luxury that is too expensive for most species. To be flippant, the test of evolutionary fitness that the human ‘big brain’ passed on the African savannah was not the ability to do calculus, but the ability to successfully function in social groups that multiplied the effectiveness of individual human capabilities through social interaction. The fact that this evolution in social cognition also left most humans capable of higher cognitive thought has obviously very recently (in evolutionary terms) had great impact, but through most of our history the value of a big brain to homo sapiens was in enabling us to work with, provide for, and understand (well, sort of) our fellow human beings.
The second assumption, which I would suggest is reasonably obvious only in hindsight, is that the most logical method of evolving the neurological capability to understand others is based on using ourselves - and therefore the neurological processes that we use to operate ourselves - as a model and a base. It is presumably far easier to evolve and adapt the capability to use our own neural network to model the behaviour of others than to build an entirely new and independent neurological network to do the same job. In addition, our existing neurology presumably works reasonably well at explaining human behaviour (at least for a sample of one). An entirely new and independent system using a different neurological construct might not be as accurate in predicting the outcomes that are generated by the original system.
Third, human development and maturation is a process in which certain steps lay the foundation for further development, and altered and/or dysfunctional development in one area may have downstream consequences in other areas, even when those areas may not be closely linked after maturation.
Fourth, while the discussion below describes various facets of autism, there is no implication intended that there is one ‘autism’ or that all autistics have the same set of capabilities, issues or symptoms. I have often seen attempted refutations of autism findings or hypotheses on the basis that they are not ‘universal’ or that other disorders also share some of the same characteristics. In the discussion below I fully acknowledge that there are both variations and exceptions in autistic capabilities. Admittedly my exposure to the scientific literature is limited, as is my personal experience with autistic children, but I am a proponent of the concept of multiple etiologies (one that I believe is fairly widely shared), “where multiple pathways funnel through final common pathways, while one or a few mechanisms lead to multiple consequences” (Herbert, 2005). As such, I do not see the absence of ‘universality’ as necessarily implying anything other than that variations in impact on certain neurological systems exist due to variations in etiology and developmental environment, among other factors.
Hypothesis – ACC and AI Dysfunction and ASD
Belmonte and Carper (1998) wrote that:
"the effect of a lesion in a developing brain often is more severe than or is distinct from the effect of a similar lesion in a mature brain, because brain systems downstream from the lesion may depend on correct input for their proper maturation."
I would hypothesize that ASD results from early ACC and AI dysfunction in a susceptible pre-autistic brain. A common thread in ASD is sensory integration issues (see below). Building on Posner and Rothbart’s hypothesis, a disruption in the brain’s ability to integrate sensory input and integrate autonomic functions with interoception could impact the development of executive functions, as well as corresponding development within the frontopolar cortex and VMPFC, and ultimately have far reaching developmental consequences, including emotional and social development, and result in dysfunction within the mentalizing and mirror neuron systems.
This disruption may be the result of dysfunctional ACC and AI input, VEN dysfunction, or both (altered ACC and AI functioning might affect the VEN developmental and migrational pathways). ACC and AI dysfunction could also be totally unrelated to VEN issues and still have the same result. But given Allman et al (2005)’s findings of VEN issues in ASD, and the roles of the ACC and AI in neurological development and functioning, VENs are a common thread linking these areas together and a logical candidate for involvement.
The result, given the roles of the ACC, AI and frontopolar cortex discussed above, could be the development of ASD in an otherwise pre-autistic but susceptible brain.
Sensory Issues, Interoception and Attention
Sensory integration issues in ASD are widely (anecdotally) reported but (as far as I can tell) largely unquantified. An exception is Kern et al (2007), which examined the relationship between auditory, visual, touch, and oral sensory dysfunction in autism in 104 children and adults. The authors indicated that findings suggested that all main sensory systems and multi-sensory processing appear to be affected; that sensory processing dysfunction in autism is global in nature; and that sensory processing problems need to be considered part of the disorder. Among other results, the authors also found that sensory disturbance correlates with severity of autism in children, but not adolescents and adults.
Interestingly, another paper by Williams et al (2006) found a more pronounced impairment in sensory perceptual abilities in childhood in autism vs. controls, but that these impairments become attenuated, such that no significant differences were detected between autistics and controls in later adolescence and adulthood. The suggestion was that this attenuation may reflect sensory system developmental amelioration over time. The authors stated that greater sensory disturbance in children would be consistent with the findings reported in the small sensory literature of the prominence of sensory symptoms in children, plus the frequent use of sensory interventions in children for calming.
Belmonte and Carper (1998) noted studies that autistic children are universally dyspraxic (Jones and Prior, 1985), have disturbances in gait (Vilensky et al, 1981), that can persist into adulthood (Hallett et al, 1993), and motor anomalies clear enough that blind raters could identify autistic from normal children based on mobility and other factors seen in home movies from before autism was suspected (Adrien et al, 1992).
A more direct account of autistic sensory issues is that of Chandima Rajapatirana, an autistic writer, from Time magazine's cover story on Autism (May 15th, 2006 issue, written by Claudia Wallis):
"[The] knack of knowing where my body is does not come easy for me. Interestingly I do not know if I am sitting or standing. I am not aware of my body unless it is touching something... Your hand on mine lets me know where my hand is. Jarring my legs by walking tells me I am alive."
Chandima Rajapatirana's account may not be typical, but it is illustrative of the extent to which sensory issues may impact some autistics. But these issues are not part of the DSM-IV, and this potentially affects the amount of attention that they receive in studies of ASD.
The minicolumn hypothesis already suggests the potential for an increased number of minicolumns per macrocolumn, which could result in an amplification of thalamic input to the ACC and AI. In addition, smaller neurons, especially in Layer V, would reduce the potential for longer distance connectivity. Factor in the potential for VEN dysfunction (potentially linked to a variety of reasons) and/or increased cerebellar input and the result of sensory integration issues could be a significant impact on ACC and AI functional development.
VENs were previously thought to appear around the fourth month after birth (Allman et al, 2001). While current findings indicate that they first appear in very small numbers in the 35th week of gestation and that about 15% of the final postnatal population are present at birth, the previous fourth month estimate suggests that VEN counts start to increase significantly about that time. Given their location, it is a logical assumption that they begin to play an increasingly significant role in AI and ACC functionality at this time, which coincides with the infant’s capacity to hold its head steady, smile spontaneously, track an object visually, and reach for that object. Allman et al (2001) suggest that VENs may participate in the neural circuitry responsible for these functions, which are related to focused attention and emotional expression. Further, some of the VENs present in four to eight month olds bear indications of potential migration, suggesting that this is an active period for VEN proliferation. It is also a period in which infants develop increasing control of their senses and bodies.
I would suggest that proper sensory integration development is required for the development of interoception, and that interoception issues would hinder efforts to control and regulate attention. As per Singer’s indication that joint ACC-AI activation may involve a simultaneous generation of both a feeling and an emotional motivation, along with its associated autonomic effects (see Anterior Insular Cortex, above) any dysfunction in this functionality in a developing infant could presumably weaken the link between emotion and associated motivations. If this link is self-reinforcing via autonomic responses feeding back into interoception, or if alterations in autonomic responses connected with emotional shifts are dysfunctional, then disruption of autonomic responses would also have an additional impact on the regulation of emotional states.
Posner and Rothbart (1998) suggest that early infant life is concerned with the regulation of state, including distress, and that attention appears to be important in developing this form of control. They state that prior to three months, holding and rocking are usually the main means of quieting an infant, but at about three months, many care-givers attempt to distract their infants by bringing their attention to other stimuli. As infants attend, they are often quieted and their distress appears to diminish. I would suggest that if this process is interrupted – via a reduction in the infant’s developing ability to subjectively recognize distress and other emotional states, or by a weakening of the distress-regulating impact of attention (e.g. reduced autonomic feedback or reduced ability to perceive the feedback via interoception) - then the link between attention and comfort would be compromised. Variations in AI or ACC development that alter the infant’s ability to either feel or indicate distress could result in a slower or weaker development of attention as a means of control. Also, disruption of the development of interoception and/or attention could also impact development of the infant’s attempts to gain more direct control of attention and become more involved in attempting to solicit adult attention, i.e. interact with the world.
This dysfunction in autistics may also partially account for the prevalence of stimming and repetitive motions in ASD. A.R. Damasio's somatic-marker hypothesis of consciousness suggests that the sensory representation of the body is the basis for the mental representation of the sentient self, allowing the brain to distinguish the inner world from the outer world. As such, stimming and repetitive motions may be an attempt to generate sensory input, to increase AI functionality and enable autistics to help make sense of their internal and external environment.
One route for the impact of interoception issues via VENs may be through the vasopressin 1a receptors. Vasopressin has a role in promoting social bonding, and any disruption in V1a receptors in the AI or ACC could have an impact. In mice, V1a receptor knockout mice have a profound deficit in social recognition (Bielsky et al, 2004). A hypothesis is that vasopressin may play a role in the formation of social bonds between caregivers and infants, and that a disruption in these bonds, combined with a weakening in the development of attention, may impede the development of joint looking and gaze following, which would have implications for the development of joint social attention (see below).
Note that this impairment would be neurologically driven – this is not the refrigerator mother hypothesis, but a more profound neurological disruption. I am also NOT suggesting that autistic children do not show attachment to their caregivers. In a meta-analysis of sixteen studies of autism and attachment by Rutgers et al (2004), a majority of studies found evidence for attachment behaviours in autistic children, although children with autism were significantly less securely attached to their parents than comparison children, and that there was a link between level of mental development and security of attachment.
A further thought on interoception is that the development of subjective self-awareness of self is presumably required before one can project this capability to the analysis of others. As such, sensory integration and interoception issues that delay development of subjective self-awareness would presumably impact the extension of this capability to others. Schultz et al (2003) cite Gusnard et al (2001) as suggesting that the medial prefrontal cortex (MPFC) is involved in any kind of thought that uses the self as a referent. And Frith and Frith (2003) indicate that the MPFC is activated when we attend to our own mental state as well as the mental state of others. If this MPFC role in subjective awareness and understanding is developmental in origin, then issues related to interoception and self-awareness could presumably impact the development of mentalizing capabilities (more below).
ACC – Amygdala Interaction
Posner and Rothbart (1998) suggested that “amygdala-cingulate interaction might be a reasonable candidate for the earliest form of self-regulation in the infant”. Evidence of dysfunction in this relationship can be extrapolated from analysis of the amygdala in those with ASD. Schumann and Amaral (2006) indicated that in typical development the amygdala undergoes a prolonged postnatal increase in volume, reaching adult size in adolescence. In ASD, the amygdala is larger in young children and reaches adult size by about 8 years old due to faster growth, but then ultimately declines in size and has fewer neurons than age-matched controls. One hypothesis is that the amygdala has a normal or even increased number of neurons in early postnatal life, but that due to stress overload and the resulting overproduction of cortisol, over time this heightened response could have damaging effects leading to the loss of neurons and a smaller amygdala. The authors indicate that anxiety is a common comorbid feature of autism. I would suggest that this is due at least in part to the reduced ability of the ACC to mediate distress.
As mentioned above, the frontopolar cortex is activated by executive functions, and overlaps with the ventromedial prefrontal cortex (VMPFC - areas 25, ventral 24, 32 and medial 10, 11 and 12), which is also implicated in some executive function tasks. Building on Posner and Rothbart (1998)’s hypothesis that executive function and control development is linked to the ACC, and Allman et al (2002)’s hypothesis of an evolutionary linkage between the ACC and area 10, a disruption in ACC and/or its links with the prefrontal cortex could impact the ability of the brain to develop the capability to direct attention, disrupting the development of follow-on capabilities, including self regulation of emotion, conflict resolution, error monitoring and correction, etc. This disruption may further carry over to impact mentalizing capabilities (see below), as performance on tests of executive function predict performance on theory of mind tests (Hill, 2004b). There may be an overlap in the areas of the brain activated by each function.
Hill (2004a) and Hill (2004b) give a good overview of executive dysfunction in autism. Autistics are not impaired on all executive function tasks. One area of impairment appears to be that of mental flexibility (Hill 2004a). Reduced mental flexibility is illustrated by perseverative, stereotyped behaviour and difficulties in the regulation and modulation of motor acts. Mental flexibility presumably activates the frontopolar cortex (Decety and Jackson, 2004), given this area’s engagement during intentional or explicit rule induction, as well as its suggested roll in mediating switching between different executive processes (Strange et al, 2001). In the WCST task, which is known to activate the frontopolar cortex, autistics usually (but not universally – Hill 2004b) experience difficulty in adapting to shifts in sorting by a new rule due to perseverative responses (Hill 2004a).
Autistics also have difficulties with some aspects of inhibition, specifically including the ability to inhibit prepotent response (Hill 2004b). A study by Stieben et al (2007) linked inhibitory control to dorsal ACC and error-related negativity (ERN) EEG activity in control children. Allman et al (2002) notes that while the error recognition and correcting function of the ACC evolved in our ancestors prior to VENs, that "spindle cells may serve to augment and relay the error-correcting information to other parts of the brain." This may include the frontopolar cortex, which also appears to play a significant role in inhibition (Decety and Jackson, 2004). One researcher (J Russell) suggested that autistic inhibitory issues may be related to the apparent arbitrary nature of the rules involved in the tests (Hill 2004b). Given that the frontopolar cortex is implicated in rule induction, dysfunctional ACC – frontopolar cortex interaction could be implicated in difficulties in deciphering arbitrary rules. Further evidence of a more direct inhibitory role of the frontopolar cortex is cited by Decety and Jackson (2004), in which several studies identified the frontopolar cortex in being involved in inhibitory or regulating processing (discussed further below).
Interestingly, Posner and Rothbart (1998) linked both the ACC’s role in error recognition and the role of inhibitory control in an experiment similar to the ‘Simon Says’ game. Children aged 40 – 48 months were asked to execute a response when given a command by a toy bear, but inhibit it when given a command by a toy elephant. Children up to 42 months scored no better than chance, but by 46 months they were virtually perfect, as inhibitory (physical) control developed. But the children who scored at chance levels apparently recognized they were not supposed to respond, as demonstrated by both slower incorrect response times and slowed responses to trials following errors. If inhibitory control is a frontopolar cortex function (discussed further, below) and error recognition is an ACC function, then this test demonstrates both the link between the two executive functions, and presumably the existence of a developmental link between the ACC and frontopolar cortex.
As per the key assumptions stated above, I would suggest that while area 10 is involved in rule induction and learning, that this is originally a socially-linked involvement, and that potentially the original ‘rules’ that train this area of the brain are both social in nature and are linked to experience-based development. In addition to area 10’s activation during emotionally charged moral dilemmas (mentioned above), further evidence supporting this involvement may come from Greene et al (2001), in which fMRIs were conducted on subjects being asked to make choices regarding moral-personal, moral-impersonal, and non-moral conditions. The medial portions of areas 9 and 10 (among other areas) showed significantly more activation during the moral-personal choices. I would suggest that this is because the role of area 10 is at least in part to decipher the social ‘rules of the game’, which may be among the most complicated and arbitrary rules in human existence. This role fits well with Allman et al (2002)’s hypothesis of a link between the ACC and area 10, the role of VEN neurotransmitter receptors in reward (dopamine) and punishment (serotonin), and Allman et al (2005)’s suggestion of one of the roles of VENs as being linked to intuition, with dysfunction in this role being a key deficit in ASD.
As suggested by Allman et al (2002), the role of the ACC may be to monitor the current state of reward and punishment and signal the need for behavioural adaptation, including to the frontopolar cortex. As such, a dysfunctional linkage between the two might impact the development of mental flexibility. Given the evidence that the frontopolar cortex is linked to rule changes and learning, while the dorsal, ventral, and dorsolateral prefrontal cortices are involved in guiding behaviour according to previously learned rules, dysfunctional ACC – frontopolar cortex connectivity could therefore have at least four major consequences: i) less frontopolar cortex experience-based development of the capacity to make adaptive responses to changing conditions, ii) greater reliance on existing learned strategies in a given situation due to reduced or altered frontopolar cortex activation, iii) a reduced ‘learned-experience’ set to call upon in a given situation, and iv) a resulting alteration in prefrontal cortex development, including potentially altered and/or dysfunctional development.
Frontopolar cortex developmental issues might therefore impact the frontal cortex as a whole. Bechara (2002) notes that impairments of emotional and social behaviour are often observed after damage to the VMPFC. "Previously well-adapted individuals become unable to observe social conventions and decide advantageously on personal matters. Their ability to express emotion and to experience feelings in appropriate social situations becomes compromised." Studies determined that theses deficits were the result of impaired ‘judgment and decision-making’. If VMPFC damage impacts the application of social knowledge, then presumably a frontopolar developmental impact on the VMFPC could be felt as an impaired ability to acquire this knowledge. Elsinger et al (2004) builds upon the notion of prefrontal cortex dysfunction from a developmental perspective, suggesting that age, experience, and an ‘altered integration and interplay of cognitive, emotional, self-regulatory, and executive/meta-cognitive deficits’ contribute to differences in behavioural outcomes.
Eye Contact and Face Processing
Averted gaze (i.e. avoiding eye contact) is a defining feature of autism. A reasonable assumption is that lack of eye contact can have an impact on the ability of autistics to decipher emotion in others. Autistics have also been reported to show less activation in the ‘fusiform face area’ (FFA) in the right fusiform gyrus (FG). This area is implicated in face perception in non-autistics (Schultz et al, 2003). Ashwin et al (2007) indicated that there is much evidence showing that autistics use a different cognitive style than controls for face processing, and found differences in brain activation during perception of fearful faces. Sasson et al (2007) identified that when processing emotional information in social scenes, individuals with ASD fixate on faces less than controls, and fail to orient to faces rapidly when facial information is available. But as Gernsbacher and Frymiare et al (2005) wrote, "It is not too surprising that autistics are less likely to activate the putative face processing area; autistics are less likely to look at faces." They indicated that autistics have described eye contact as threatening, painful, or draining. I would hypothesize that variations in gaze and face processing in those with ASD can be ultimately linked back to issues in the ACC and AI.
According to Decety and Jackson (2004), when reading emotions on the face of another one activates the same facial muscles at a sub-threshold level, triggering the same expressions on one’s own face, even in the absence of conscious recognition of the stimulus. Evidence further suggests that this results in autonomic changes, and is associated with a subjective experience of the corresponding emotion. As stated above, Singer et al (2004) suggests that the ability to understand the feelings of others and to empathize evolved out of our own ability to subjectively represent our internal body states. If so, then impairment in interoception (AI) or in the ability to generate the appropriate autonomic response (ACC) could render face emotion processing more difficult for those with ASD. In this case the face could be a less useful tool in eliciting information regarding the emotional state of others. As well, changes in gaze are presumably linked to attentional shifting and mental flexibility, which are also indicated (above) to be issues in autism, and will be explored further below (in Joint Attention).
Further, the left amygdala (among other regions) is activated during the discernment of gaze direction, and the right amygdala is activated when another individual’s gaze is directed at oneself, suggesting that the amygdala plays a role in reading social signals from the face (Kawashima et al, 1999). In much of the animal kingdom, direct gaze indicates threat. As Frith and Blakemore (2004) indicate, this is clearly not the case in humans, who use eye gaze to indicate a wide variety of positive and negative emotions and intentions. From an evolutionary perspective it makes sense that the neurological mechanisms that interpret direct gaze as a threat would continue to exist, but would be mediated by the social brain. M.A. Williams et al (2005) indicated that emotional facial expressions - especially expressions that convey potential threat - increase amygdala activation, even when the face is masked from awareness. Ashwin et al (2007) also confirmed activation of the social brain network in face processing in controls.
An amygdala subjected to less social brain mediation would presumably play an increased role in face perception, and the result might be more ‘discomfort’ in face processing, with eye gaze avoidance being an adaptive response. One source of this reduced mediation might be attention related. As mentioned above, dysfunctional ACC attentional development could have an impact on ACC regulation of the amygdala. Over time, the impact of altered ‘social brain’ development would also have an impact. Dalton et al (2005) reported that in autistic subjects, the amygdala is activated to an abnormal extent during a direct gaze upon a non-threatening face. As Richard Davidson (one of the researchers in Dalton et al, 2005) pointed out, it is over-aroused amygdalas that make autistic children want to look away: "Imagine walking through the world and interpreting every face that looks at you as a threat, even the face of your own mother." Evidence of a heightened response to direct gaze was also indicated by Kylliainen et al (2006). In measurements of skin conductance responses (SCR) to another person's gaze, ASD children showed greater SCR to straight gaze over averted gaze (i.e. a variation in autonomic response and discomfort), versus no difference in controls, indicating enhanced arousal to eye contact in those with ASD. Given the potential for both a reduced ability to read facial emotions and heightened discomfort when looking at faces, eye gaze avoidance in autistic children as an adaptive response should not be surprising.
Face processing is an emergent and developmental skill that is heavily mediated by early experience with faces (Sasson, 2006). As such, it is not unreasonable to expect differences in eye gaze behaviour to lead to differences in face processing between autistics and controls. But recent research suggests that the differences in FFA processing are not as great as once thought, while other variations in processing exist in the social brain network. Hadjikhani et al (2004) found that individuals with ASD activated the FFA and other brain areas normally involved in face processing when they viewed faces as compared to non-face stimuli. This suggests that ASD face-processing deficits are not due to a simple FFA dysfunction, but rather to more complex anomalies in the areas of the brain involved in social perception and cognition. Hadjikhani et al (2007) replicated the original findings of significant activation of face identity-processing areas (FFA and inferior occipital gyrus, IOG) in ASD, but also found activation to faces in this broader ‘social brain’ network. They identified hypoactivation in the more widely distributed network of brain areas involved in face processing, including the right amygdala, inferior frontal cortex (IFC), superior temporal sulcus (STS), and face-related somatosensory and premotor cortex, with atypical patterns of activation in the IFC and STS, suggesting that areas belonging to the mirror neuron system are involved in the face-processing disturbances in ASD.
Pierce et al (2004) looked at responses in adult autistics to both personally meaningful faces as well as to those of strangers. They also found significant FFA activity in autistic subjects, as well as greater FFA activity in response to familiar over stranger faces, plus right hemisphere dominance in response to both face types. The authors suggest that the study design may have contributed in part to the greater than expected activation to the faces of strangers, in that the use of familiar faces may have heightened the overall level of interest. They suggested that FFA activation may be related in part to neural systems related to social drive and motivation or cognitive and attentional engagement, reflecting (as above) that atypical FFA results in ASD may be related to the systems that modulate FFA activity.
A further interpretation of FFA activation is discussed in Schultz et al (2003). This paper noted Gauthier’s finding that the FFA responds preferentially to any class of object for which a person is perceptually an expert, which suggests that the FFA is organized by experience. Further, an experience bias normally indicates that it is important to discriminate faces by identity, biasing the FFA toward individual identification. The FG also encodes semantic information for category identification (including anything that helps identify faces as a category), and Schultz et al (2003) suggested that social information would also be stored, since the social retrieval of faces and social judgment is a repeated perceptual experience. The suggestion is that the middle FG area would store general information about people or a meta-representation of ‘peopleness’, including facial information during social events (i.e. facial affect), in an area overlapping with the FFA. Schultz et al (2003) also indicated that the FFA is activated as part of the social network, contributing to the identification of social events and emotions related to faces. The implication was that deficits in FFA processing in ASD would also impact social cognition. But if the FFA is activated in ASD then potentially this FFA social expression recognition functionality may also be active too.
Another area preferentially but not consistently linked to face processing is the superior temporal sulcus (STS). The STS is activated when observing the behaviour and also when retrieving information about the behaviour of living things, and perhaps with any complex behaviour. This area is also part of the mentalizing network of the brain (Frith & Frith, 2003). Hoffman and Haxby (2000) indicate that the STS has a role in the perception of eye and mouth movement and the changeable aspects of faces (e.g. expression), as well as in determination of eye gaze direction (see Joint Attention below). Blair (2003) indicates that the FFA is probably more associated with facial recognition, while the STS is more involved in processing social communications, presumably including recognition of affect. Even so, both the STS and FFA have a greater response to emotional over neutral expressions, and tasks requiring increased attention to emotional expressions result in a greater activation of both. Blair (2003) indicated that social expressions activated both the FFA and STS, and also indicated that STS can be further activated during recognition of emotion after preliminary activation, possibly as a consequence of amygdala activity. In effect, this subsequent activation may be an interpretive activity as part of this area’s role in the social brain network.
Clear evidence exists that autistics can recognize emotional states. Back et al (2007), for example, found that autistics were able to infer mental states from dynamic and static facial stimuli at a better than chance rate (but less accurately than controls). Autistics were also as successful as controls in recognizing mental states when eyes were presented in isolation or in the context of the whole face (presumably a function of STS processing). While studies have reported that autistic children have difficulty recognizing the emotional expression of others, Blair (2003) indicates that these studies have failed to match children on mental age. When appropriately matched, he states that autistic children have usually been found to be unimpaired in facial affect recognition. Further, several studies have found the emotion processing impairment to be pronounced only in complex cognitive emotions such as surprise or embarrassment. The issue in autism may not be an inability to recognize affect (when the autistic is looking at the face to be ‘read’). The evidence above indicates that autistics can use the FFA and STS for face processing and recognition of affect. Instead, the difficulty for the autistic, given reduced activation of the social brain network, would be in processing the emotion and (as per Blair, 2003) representing the emoter’s intent (see Emotion and Empathy, below). This is how I would interpret the findings of Hadjikhani et al (2007) of a hypoactivation (as distinct from lack of activation) in the more widely distributed network of brain areas involved in face processing, including the STS.
Interestingly, Pierce et al (2004) also noted significant amygdala activation in both controls and autistics, unlike Hadjikhani et al (2007)’s finding of reduced ASD activation. Given the conflicting amygdala results, I would interpret amygdala activity as changing over time. As noted above, Dalton et al (2005) showed heightened amygdala activity in children, while Hadjikhani et al (2007) and Pierce et al (2004) both examined older autistics. In addition, Pierce et al (2004) included a significant number of familiar faces, which could have motivated a higher emotional response among autistics vs. Hadjikhani et al (2007). As noted in Schumann and Amaral (2006) above, the amygdala in autistics reaches adult size by 8 years old and then ultimately declines in size. I would suggest a) that over time the role of the amygdala changes with autistic neural development, and b) the later development of some social brain capabilities may impact key developmental windows of other functions, reducing some downstream capabilities. As an example, the later development of theory of mind capabilities may miss the developmental window that allows for effortless social brain mediation of amygdala reactions to faces.
Emotion and Empathy
A significant consequence of interoception issues may be the inability to fully understand one’s own emotional state, and the emotional state of others, which could manifest as a difficulty in developing empathy.
Decety and Jackson (2004) defines empathy as "a complex form of psychological inference in which observation, memory, knowledge, and reasoning are combined to yield insights into the thoughts and feelings of others". Three primary components are involved: a) an affective response to another person - often but not always including sharing their emotional state – based on perception-action coupling (see below) that leads to shared representations of behaviour, b) self-other awareness, separating temporary identification with the other from emotional contagion, and c) the mental flexibility to adopt the subjective perspective of the other but subject to self-regulation (Decety and Jackson, 2004).
Shared representations between self and others is based on the concept that the perception of emotion in another automatically activates the neural mechanisms that are responsible for generating emotions in us, leading to our own representation of that behaviour. The underlying theory is that perception and action are linked – perception is a means to action and action is a means to perception – and that this is hardwired into our nervous system. In this hypothesis, the recognition of emotion in others (especially the intuitive recognition of same) stems from at least partially recreating the bodily and motor components of that emotion in ourselves. This is supported in part by the finding in lesion studies of paired deficits between emotional production and emotion recognition. In this view, some developmental psychologists propose that the understanding of others is ‘primarily a form of embodied practice’. "Humans develop and maintain their self-concept through the process of taking action and then reflecting on what they have done – that is, the sensory consequences of their actions – and later in life, what others tell about what they have done." From this, "the understanding of the other person emerges in part from being ‘like them’ in action, through imitation, and that this provides the basic mechanisms for empathy." (Decety and Jackson, 2004)
Carr et al (2003) indicated that the AI plays a fundamental role in the recognition of emotion in others, relaying action representation information to areas processing emotional content. They demonstrated that both the observation and imitation of emotional facial expressions in others activated overlapping neural networks. This included the pre-supplementary motor area and the rostral cingulate zone (RCZp, part of the ACC), causing a robust pre-motor response, supporting the concept that recognition of emotion in others is linked to sub-threshold action representation and subsequent interpretation within the self as per Decety and Jackson (2004), and Singer et al (2004). But if interoception issues impact the effective subjective understanding of oneself, then reading a representation of others within ourselves would also be problematic.
Self-awareness and other-awareness are also potentially linked in another key manner. It has been suggested that both develop in close synchrony during the 2nd year because both types of cognition are based on a capacity for secondary representation. Self-awareness is based on secondary representation because the self as an object of knowledge is a secondary representation. Other-awareness requires a secondary representation as it implies taking the perspective of another person into account. The development of this capability may be the driver of significant increases in social-cognitive competence during this period (Decety and Jackson, 2004), e.g. triadic joint attention, theory of mind, empathy, etc.
This development may be linked to executive function. Carlson and Moses (2001) demonstrated that inhibitory control was strongly related ToM performance in preschool age children. A conflict task, requiring a novel response in the face of a conflicting prepotent response, was significantly related to ToM. The authors indicate that the findings suggest that inhibitory control may be a crucial enabling factor for ToM development, possibly affecting both the emergence and the expression of mental state knowledge. Kelly et al (2004) conditionally linked inhibitory control to the frontopolar cortex.
Decety and Jackson (2004) links empathy to perspective taking, i.e. the ability to more or less consciously adopt the subjective point of view of another. They stated that perspective taking develops gradually, and that it is around 18 months that children demonstrate an emerging awareness of the subjectivity of other people’s emotions. They also note that realizing that others can have a perspective that differs from one’s own does not mean that one is able to adopt the other’s perspective. People are fundamentally egocentric and their predictions of the feelings of others tend to be largely based on their predictions of how they would feel in the same situation. This is compatible with simulation theory, in which one is assumed to predict the behavior and mental state of others by simulating it as if we were in the same situation. Children especially often have to judge situations from a first-person perspective, and one study found that they were better at predicting what another would see if they had actually been in that situation themselves first. As such, self-perspective is the default mode of the mind.
Mental flexibility and self-regulation are therefore important parts of empathy in that they enable one to regulate one’s own perspective, recognizing that the other is like the self but while maintaining a clear separation between self and other. In this Decety and Jackson (2004) suggested that evidence from clinical neuropsychology and neuroscience points to the frontopolar cortex as being chiefly involved in inhibitory or regulating processing. Evidence suggests that damage to this area results in individuals taking an excessively egocentric perspective in moral dilemmas, revealing a lack of inhibition (or modulation) of self-perspective at the conceptual level. Decety and Jackson (2004) also cited neuroimaging studies of perspective taking supporting the hypothesis of an inhibitory role of the frontopolar cortex for adopting the subjective viewpoint of others, whether the shared activated representations are motor, conceptual, or emotional in nature. Further support for this role was provided by fMRI studies in which involvement of the right lateral prefrontal cortex was detected when participants inhibited a prepotent response in a sensory motor task and also in a deductive-reasoning task.
As such, Decety and Jackson (2004) argues that the inhibitory component of the frontopolar cortex is required to regulate and tone down self-perspective to allow for the evaluation of the other-perspective. The authors indicate that this is required because the automatic link between perception and action renders the self-perspective as the default mode and prepotent response. Regulation of this prepotent response (potentially an executive function impairment in autism, as discussed above) allows for cognitive and affective flexibility. I would suggest that dysfunction in ACC-frontopolar cortex linkages, potentially related to issues of VEN functionality, may drive the impairment in inhibitory control in ASD.
The above is not to say autistics have universal impairment in empathy (see Key Assumption 4, above). Gernsbacher (2006) discusses the example of Walker Hughes, an autistic who clearly demonstrates a capacity for empathy. I would also suggest that difficulties in the ability of autistics to empathize may too often be inaccurately linked in the minds of others to a lack of desire to empathize or emotionally bond with others. Difficulty in understanding others does not automatically equate to a lack of desire for understanding or a lack of desire to try. But that is another post.
Joint attention can be defined as the interaction between two people about a third object (Frith and Blakemore, 2004). But as per Frith and Frith (2003), joint attention can be defined according to strict or lenient criteria. If joint looking or gaze following is required, then this can occur relatively early. From approximately 12 months, infants tend to automatically look at a target that an adult is looking at. But this only happens when the target is already within the infant’s point of view. Strictly defined, joint attention indicates an implicit awareness that others can pay attention to different things than oneself, and that others’ attention can be directed to coincide with one’s own interests. This level of joint attention emerges around 18 months, and indicates an implicit ability to mentalize (Frith and Frith, 2003). Joint attention issues are probably universal in autism (Hill, 2004a). I would suggest though that joint attention difficulties are a follow-on consequence of some of the issues discussed above.
Infants can reflexively use movement of gaze as a priming cue for own eye movement at three months. But as Frith and Frith (2003) indicate, this is probably both innate and rests on an entirely different set of neural capabilities than those demonstrated by joint attention. Voluntary gaze following is first seen around 12 months, and suggests a primitive understanding that gaze involves a relation between a person and the object of their gaze. Note that this is not the same as joint attention, strictly defined, and this voluntary gaze following occurs only when the target object is within the infant’s line of vision (Frith and Frith, 2003). Another similar activity is ‘social referencing’, which starts at around 8 to 10 months, and involves an active attempt by infants to obtain emotional cues from others to assist in their assessment of uncertain or ambiguous objects or situations (Decety and Jackson, 2004; Blair, 2003). Both cases require the infant to actively focus attention on another, and I would suggest that both could be an issue for infants who will later be diagnosed with ASD.
One could speculate, as per the discussion in Eye Contact and Face Processing above, that if children with ASD have difficulty in looking at the eyes of others then gaze following could also potentially be impacted. One of the areas linked to gaze following is the superior temporal sulcus (STS). In a test of congruent vs. non-congruent eye gaze, Pelphrey et al (2005) found that both controls and autistics activated the STS region in response to shifts in eye gaze. But only controls showed STS activity that differentiated congruent vs. incongruent gaze shifts. In autistics they reported that the only reliable changes linked with incongruent gaze was activity in the insular cortex and inferior frontal gyrus. In controls, greater STS activity was associated with violation of expectations and a need for revision of expectations. In autistics, incongruent activity did trigger a response, so recognition of incongruent activity did occur. But the response did not involve activation of the social network as per the controls. From this one could suggest that ASD issues with joint attention are related not to gaze identification (which is presumably a precondition of same) but to 'attention' and the subsequent analysis of the gaze intention of others, i.e. mentalizing.
According to Carpenter et al (2002), typical infant development follows a reliable sequence. Infants first share others’ attention (alternating gaze between an object and an adult), then follow others’ attention and then behaviour (imitation), and then they direct others’ attention (declarative gestures) and then behaviour (imperative gestures). Autistic children tend to follow the typical pattern of sharing, then following, then directing, but with attention following rather than leading behaviour – i.e. follow behaviour, share attention, direct behaviour, follow attention, and then direct attention (Carpenter et al, 2002).
This makes sense if one considers that autistics may have an early difficulty in the development of an understanding of others, due to both attentional issues and difficulties in interoception and the development of self-understanding. Hobson and Hobson (2007) suggest that “the propensity to adopt the bodily anchored psychological stance of another person is essential to certain forms of joint attention and imitation, and that a weak tendency to identify with others is pivotal for the developmental psychopathology of autism”. But if the understanding of others is ‘primarily a form of embodied practice’ (Decety and Jackson, 2004), then how does one accomplish this in the right order if one has difficulty with interoception and in embodying one’s own persona?
Identification with others could be also potentially be impacted via issues in the detection of external agency. Blair et al (2002) indicated that there is some evidence of a reduced sensitivity to agency cues in ASD, especially related to movement. Evidence suggests that the inferior parietal cortex plays a role in the elaboration of the image of the body in space and time, and that the right inferior parietal cortex in conjunction with the prefrontal areas may be critical in distinguishing self from others (Decety and Jackson, 2004). In a shared representation framework, an efficient body schema may be necessary not only for recognizing one’s own actions but also for subjectively understanding the actions of others. Decety and Jackson (2004) suggests that the inferior parietal cortex in conjunction with the prefrontal cortex plays a pivotal role in the sense of self by comparing the source of sensory signals, crucial in maintaining a distinction between the self and the other and keeping track of the origin of feelings.
Farrer and Frith (2002) suggested an AI role in this process, concerned with the integration of input associated with voluntary movements, while the inferior parietal cortex represents movements that can be applied to the actions of others as well as oneself. Being aware of causing an action was associated with activation in the AI, while being aware of not causing the action and attributing it to another was associated with activation in the inferior parietal cortex. Farrer et al (2003) took this further, suggesting that the sense of agency is based on a continuous mechanism, with greater correspondence of input indicating a greater sense of agency and greater insula activity, and weaker correspondence resulting in higher levels of activity in the right inferior parietal lobe. I would suggest that AI-based interoception issues would have a significant impact on this mechanism: if one has difficulty in detecting and determining the origin of one's own sensory input, then how can the brain use this information to determine the degree of correspondence of input required to attribute agency?
To the extent that inferior parietal lobe development depends on AI functioning and connectivity, inferior parietal lobe functionality could be impacted by AI dysfunction. I would suggest that Pelphrey et al (2005)’s finding of insular cortex activity but not further activity in the mentalizing system in those with ASD in incongruent gaze detection is a demonstration of the inability of the AI-inferior parietal cortex agency determination mechanism. I would suggest that information processing could not get past the agency determination 'gate' to activate the mentalizing system. As such, AI issues could explain issues related to social referencing and gaze following, both of which presumably require a functional sense of agency attribution. To the extent that they are developmentally related, they could also impact joint attention, the development of which also presumably requires the ability to assign agency.
As mentioned previously, the frontopolar cortex and cingulate cortex are involved in joint attention, along with the ventromedial frontal cortex (Williams et al, 2005). Dysfunctional connectivity between the ACC and frontopolar cortex could presumably impact the development of the frontopolar cortex and any related functionality. Strange et al (2001) suggested that the dorsal, ventral, and dorsolateral prefrontal cortices are involved in guiding behaviour according to previously learned rules. Given the suggested role of the frontopolar cortex in deciphering ‘the rules’ for later application by other areas of the prefrontal cortex, any frontopolar cortex issues would presumably have a knock on effect on development on those areas.
Given the role of the frontopolar cortex in inhibitory control (discussed above), which - as per Decety and Jackson (2004) - appears to be required to regulate and tone down self-perspective to allow for the evaluation of the other-perspective, a deficit in the development of inhibitory control may impact attention shifting required for social referencing, joint gaze, and the development of joint attention, above and beyond a weak tendency to identify with others. It may also impact the ability to develop a capacity for secondary representation that is required for both self-awareness and other awareness (i.e. the ability to see both self and other as an object of knowledge), given that this requires the development of inhibitory control.
Mirror Neuron System
One of the theories of autism is the ‘broken mirrors’ hypothesis. As stated by Hamilton (in press), "At its simplest, the broken mirror hypothesis claims that children with autism have a dysfunction of the mirror neuron system, and that this is the primary cause of their social disability (Dapretto et al., 2006; Iacoboni & Dapretto, 2006; Ramachandran & Oberman, 2006; Williams et al., 2001)." To examine this hypothesis, it is necessary first to define the mirror neuron system (MNS).
The core of the MNS is the inferior parietal lobule (IPL) and inferior frontal gyrus (IFG), which respond when hand actions are performed, imagined, planned, and imitated, and are widely assumed to contain ‘mirror neurons’ similar to those studied in equivalent regions of the macaque brain. Associated with this area is a region stretching from he lateral occipital sulcus through the middle temporal gyrus to the STS, labeled by Hamilton for simplicity as the MTG, which is robustly engaged in action observation tasks and modulated by motor performance (Hamilton, in press).
First, one of my own biases (that may reflect my lack of knowledge) is that I don’t believe that there are ‘mirror neurons’ in humans, but rather neurons that are activated as part of the MNS. To the best of my knowledge, no one has been able to clearly identify a ‘mirror neuron’ as a unique cell type, but instead, a ‘mirror neuron’ appears to be a neuron invoked during both self and ‘other’ representation. Second, as Hamilton (in press) identifies, the MNS is part of the motor system, and is an essential part of the neurological control of our own actions, in addition to its activation during imitation and action observation tasks. As per my stated assumptions (earlier), in hindsight it makes sense that rather than recreate a new system to explain others, it is more feasible that evolution has endowed humans with the ability to explain others using the same neurology that is used to operate and explain ourselves. Thus, if autistics have a dysfunctional MNS then the result should be impairment in all cognitive tasks that utilize this system, i.e. performing goal directed actions, imitating others, and understanding the goals of others (Hamilton, in press).
Frith and Frith (2003) identified the principle of rationality. Between 9 and 12 months, infants develop the expectation that agents will approach a goal in the most economical way, and are surprised if an agent does not do so, but – for example – jumps over an invisible hurdle. This demonstrates that infants can separately represent the goals of agents and the means used to reach the goal, both of which the authors indicate are likely to be important prerequisites of the ability to represent intentions. Carpenter et al (2001) studied the understanding of others’ intentions in 2.5 – 5 year old autistics vs. controls with other developmental delays using a test of others’ unfulfilled intentions in an imitation context. They found no significant between-group differences on any measure involving the understanding of others’ intentions. They suggested that those with ASD may have a slightly less complex understanding of others’ intentions than do other children (I would suggest that this might be experience-related) but was clear that any deficits in this area were not as marked as those found in ToM tests.
Hamilton et al (2007) also tested ASD children vs. controls with the same verbal mental age on four action representation tasks, as well as a series of ToM tests. ASD children demonstrated the same tendency as controls to imitate an examiner’s goals, imitate hand-goals in mirror fashion, and imitate grasps in a motor planning task, and on a gesture recognition task outperformed the controls, despite impairment on the ToM tasks. The authors concluded that autistic children have no difficulties understanding the meaning of an action or imitating the goal of an action, demonstrating clear evidence against a general imitation impairment or global MNS deficit, and therefore refuting the ‘broken mirror’ hypothesis. Instead, Hamilton (in press) identifies that where imitation errors do occur in ASD, they tend to be tasks requiring mimicry of style or meaningless actions. These types of tasks tend to require copying the low level kinematic features of an action. In contrast, goal oriented emulation tasks require the understanding the goal, from whence one can, if one chooses, plan or reconstruct the action by one’s own means, and this capability is intact in autistics.
The differences in capabilities relate to differences in the three components of the MNS, and the information flow between them. The MTG provides a visual representation of the low level, kinematic parameters of the observed action, the IPL provides a more abstract representation of the goal of the observed action, and the IFG provides a motor representation of the observed kinematic parameters, in preparation for imitating the action. Together the three nodes enable humans to plan and perform complex visually guided hand actions, to imitate another person’s action, and to understand the meaning of that action (Hamilton, in press).
In addition, the pathways between the three nodes explains differences in processing (Hamilton, in press). The primary MNS route is the EP route, in which imitation occurs in two stages. First, in the E route the visual representation of the observed action in MTG is used to infer the goal or meaning of the action in IPL. Once the goal representation is obtained, the P route between the IPL and IFG is used to plan an action based on that goal, with the planned action represented in the IFG in terms of its motor parameters. The result is a separation between action understanding and action planning. Together they become goal-emulation. Hamilton (in press) indicates however that there is evidence of another pathway, the M route between MTG and IFG that allows for the mimicking of the low level kinematic features of observed actions without requiring the prior understanding of a goal.
In typical individuals Hamilton (in press) suggests that both routes may be used separately or often together, but in ASD the suggestion is that the EP route is functional while M route may be compromised. The result is that ASD children can emulate an observed action if they understand the goal, but the ability to spontaneously imitate a meaningless gesture or facial expression may be reduced.
Hamilton (in press) raises the further hypothesis that the dysfunction is not with the M route, which may in fact be intact, but with modulating the route and deciding who and when to mimic, with this impairment being driven by dysfunction in the mentalizing network of the brain. As such, dysfunction in the ‘theory of mind network’ would be a cause rather than a consequence of abnormalities in the direct M route. I will discuss mentalizing further in the following section.
I would also suggest that this dysfunction may be related to an issue in the detection of agency. To use the EP-M model, for the person’s own system to become a mirror system it would first need the capability to detect agency, both externally (recognizing others as actors) and internally (i.e. recognizing whether represented actions are ‘self’ or ‘other’ generated). From an external perspective, children with ASD have the ability to detect agency in external actors and thus according to the EP model can use mirror neuron system capabilities to determine and emulate goal linked behaviour, assuming that attention is directed towards the ‘other’.
For the M route though, detection of external agency may not be enough, especially as some mimicry behaviour is not consciously directed, and some behaviour is ignored. As Kilner et al (2006) "everyday we are in situations in which we observe many people moving simultaneously and it seems highly unlikely that the ‘mirror system’ is activated equally by all the observed movements." Their study determined that modulation of parietal-occipital α-frequency range oscillations (which are attenuated during action observations) varied depending on whether the observed actor was facing towards or away from the subject. Given that the STS detects biological motion, the authors suggested that input to the ‘mirror system’ is filtered by modulating visuospatial attention, and that social relevance might modulate information entering the ‘mirror system’.
The M route is a direct route between observation and motor kinetics. Given the perception-action link, the M route could be assumed to require the ability to separate representation of self vs. other actions, i.e. to determine the agency associated with the internal representation of action. As per the discussion in Joint Attention (above), Frith and Frith (2002) indicated AI involvement along with the inferior parietal cortex in determining a sense of agency. Farrer et al (2003)’s suggestion that the sense of agency is based on a continuous mechanism - greater correspondence of input indicating a greater sense of agency and greater insula activity, and weaker correspondence resulting in higher levels of activity in the right inferior parietal lobe – further suggests that AI interoception issues would impact the ability to determine agency. I would suggest that this determination of agency would be an essential precondition or gate for the activation of the motor system, and thus the inability to internally determine agency would result in the exclusion of ‘other’ representations from the M route. I would go further and suggest that inability to determine agency may also impact self generated attempts to activate the motor systems and may therefore partially explain bodily control and movement issues in some with ASD. But that is another post.
Mentalizing (Theory of Mind)
Frith and Frith (2003) indicate that the mentalizing system of the brain is probably operational from about 18 months, allowing the implicit attribution of intentions and other mental states. Another major leap in development occurs between ages 4 and 6, during which explicit mentalizing becomes possible, which includes a full and explicit awareness of others’ mental states and their role in the explanation and prediction of their behaviour.
"Tentatively, we can conclude that an implicit version of the intentional stance emerges first, concerned with desires, goals and intentions. This is usually dated at around 18 months. At 18–24 months there is a convergence of several important developmental milestones, including a true understanding of joint attention, deliberate imitation and the ability to track a speaker’s intention while learning words. There is also evidence for the ability to understand knowing and seeing at an implicit level, and possibly even an implicit understanding of false belief." (Frith and Frith, 2003)
Mentalizing capabilities are widely accepted as being impaired in those with ASD. To be clear, I fully accept that those with ASD can be fully capable of performing tasks requiring explicit mentalizing (and even that degree of waffle is due only to my inability to state that ALL autistics have this capability, any more than I can state that all people have ANY particular capability). Frith (2004) references Happe (1994) as indicating that children with autism who have sufficient verbal ability to follow mentalizing tests show a delay in this area of about five years. What those with ASD appear to lack is the ability to intuitively process the intent of others. As Frith and Blakemore (2004) indicate, the slow acquisition of an explicit theory of mind does not replace the missing intuitive capabilities, and even very capable adults with Asperger syndrome demonstrate slow and error prone responses during mentalizing tasks.
Mentalizing appears to be based on three consistently activated areas of the brain – the medial prefrontal cortex (MPFC); temporal poles, bilaterally, with greater activation on the left; and posterior STS, bilaterally, with greater activation on the right (Frith and Frith, 2003). The STS is activated by biological motion and the behaviour of living things, including gaze detection. As discussed above (Joint Attention), autistics can detect gaze direction and incongruent gazes, and anyone who has spent any time with an autistic knows that they can detect biological agency. Presumably this means that the STS is functional in autism.
The temporal poles are activated in studies of language and semantics (left), semantics, autobiographical memory, and the recognition of familiar faces. Frith and Frith (2003) suggest that the region generates a wider semantic and emotional context for material being processed, on the basis of past experience, which would aid in interpretation. Included in this is the generation and retrieval of scripts, which capture the actions and goals in a particular activity or setting and time, and are built through experience. Identifying the appropriate script for a situation helps in predicting the actions of others, and provides a framework in which mentalizing can operate. Further, mentalizing may be required to understand deviations from scripts. Given that autistics can apply scripts – indeed, it may be when life deviates from scripts that autistics encounter difficulties – one could speculate that temporal pole functionality may be relatively intact in those with ASD.
Which brings us to the MPFC. In a comparison of 12 mentalizing tasks analyzed via fMRI, Frith and Frith (2003) reported that all twelve tasks activated the MPFC. The mentalizing area in question includes the most anterior part of the paracingulate cortex (BA 32) and overlaps with (depending on the nomenclature used) the anterior rostral cingulate zone and the emotional division of the ACC. BA 32 is described as a cingulo-frontal transition area that is anatomically and potentially functionally distinct from the ACC (Devinsky et al, 1995, Frith and Frith, 2003). In two of the mentalizing tasks compared above (an economic game of mutual cooperation and ‘Stone-Paper-Scissors), the MPFC was the only region differentially activated when subjects thought they were playing another person rather than a computer. This suggests that the temporal poles and STS are more concerned with the sensory signals that initiate the mentalizing process, while the MPFC is engaged by attending to the mental state of the self and others (Frith and Frith, 2003).
As with emotion and empathy and the mirror neuron system, it appears that the capacity for mentalizing about others arises from the activation of the same area of our brain that is activated in mentalizing tasks in which the subject is oneself. While the AI appears to contain the subjective representation of our body state, the MPFC contains the subjective representation of our mental state. The ability to understand oneself relies on the ability to generate a second-order or subjective representation. The first-order representation is the actual physical stimulus, while second-order representation is the mental attitudes that we take towards the stimulus. As Frith and Frith (2003) wrote, these second-order representations are decoupled from the physical world, and are no longer subject to normal input-output relationships - e.g. a change in stimulus does not necessarily correspond to a change in one’s perception of an event.
Activity in the MPFC is linked to the creation of decoupled representations of beliefs about the world that may or may not correspond with the actual state of the world. Thoughts, feelings and beliefs need not be accurate representations of reality. And as Frith and Frith (2003) indicate, mentalizing is not only about representing our own mental states as distinct from reality, but also about representing the mental states of others. Schultz et al 2003 quotes Gusnard et al. (2001) as suggesting that the dorsal MPFC is involved in any kind of thought that uses the self as a referent, e.g. theorizing about others’ minds but with explicit reference to the participant’s own frame of reference as to how they would feel in a similar situation. Decety and Jackson (2004) also indicates that the MPFC is involved in both self-perspective and other-perspective. Interestingly, self-perspectives also activated the AI and secondary somatosensory areas, while the adopting of other-perspectives also activated the frontopolar cortex and posterior cingulate.
The role of the frontopolar cortex may be of critical importance in mentalizing. As mentioned above, Carlson and Moses (2001) demonstrated that inhibitory control was strongly related ToM performance in preschool age children, and suggested that inhibitory control may be a crucial enabling factor for ToM development, possibly affecting both the emergence and the expression of mental state knowledge. Kelly et al (2004) and Decety and Jackson (2004) both linked inhibitory control to the frontopolar cortex.
Given that self-awareness and other-awareness both activate the MPFC, as per Decety and Jackson (2004), there is a requirement for regulation and inhibitory control to be able to separate the self from the other. As per above (Emotion and Empathy), this is required to regulate and tone down self-perspective to allow for the evaluation of the other-perspective. They argue that this is required because the automatic link between perception and action renders the self-perspective as the default mode and prepotent response. While the authors were considering this requirement in relation to emotion and empathy, logically it should apply to any representation that could have a self or external origin. As such, I would suggest that the same frontopolar based inhibitory control mechanism that enables one to take an ‘other-perspective’ from an emotional standpoint is also required to do the same from a cognitive perspective. As above, I would suggest that dysfunction in ACC-frontopolar cortex linkages, potentially related to issues of VEN functionality, may drive the impairment in inhibitory control in ASD. This would also impact the intuitive ability to mentalize.
Given that much of learning is both experiential and socially based, difficulties in acquiring the ability to interpret and understand the thoughts of others would have profound consequences for overall development in autistic children. Lack of understanding of others could significantly impair the development of social competencies, as well as impact the ability to learn from others. This could account for many of the criteria that lead to a diagnosis of autism, as well as secondary (non-DSM IV) criteria such as lower IQ, which in many cases may be related more to perceptual issues and the ability to learn from others than a reflection of any underlying cognitive deficits.
Language acquisition may be another area that is impacted. Frith (2001) referred to an Baron-Cohen et al (1997)’s use of the discrepant looking paradigm, in which the speaker and listener attend to different objects while the speaker utters a new word. Controls mapped the word to the object the speaker was looking at, while children with ASD erroneously mapped the word to the object that they were looking at. Frith and Frith (2003) suggest that the rapid learning of language from around 18 months may result from a child’s ability to track a speaker’s intentions when they utter a word. Carpenter et al (2002) suggested that autistic children instead used imitation (imitative learning – i.e. goal directed learning) to acquire language prior to the development of joint attention, which could account for some of the atypical features of the language of autistic children.
Mentalizing difficulties would especially impact role reversal imitation. The ability to reverse roles may require a holistic understanding of a shared interaction, with each person’s role being interchangeable, i.e. a second-order representation. Carpenter et al (2005) indicate that this ability may be a prerequisite for the learning of bidirectional communicative symbols, e.g. the understanding and use of pronouns. The authors indicate that children see themselves and others as interchangeable (I am like them and they are like me) and actively use this understanding, learning from others (self-self role reversal – e.g. you touch your hand, I touch mine), reciprocating (other-other role reversal – e.g. you pat my head, I pat yours) and engaging in collaboration through object-mediated role reversals (e.g. I hold a basket, you put items into it, then reverse). These reversal imitations become acts of cultural learning and participation. Carpenter et al (2005) suggest that the difficulties that autistic children have with role reversal at each level has a cascading detrimental impact on their cultural involvement.
The Kitchen Sink
In a post that has probably unwisely included everything but the kitchen sink, here goes. One of the points that caught my attention while reading Devinsky et al (1995) was related to a link between ACC seizures and muscle tone. Anecdotally speaking, I’ve heard a lot of reports of hypotonia in autistic children, and Ming et al (2007) identified this as the most common motor symptom (51%) in a study of 154 children with ASD. Other conditions found included motor apraxia (34%), and gross motor delay (9%). The authors noted that all of these conditions appeared to improve over time. While I don’t have corresponding data for the general population, I will speculate that these numbers are significantly higher than normally found, and thus may be more than comorbidities.
Given Devinsky et al (1995)’s linking of the ACC to early premotor processing, I would speculate that ACC issues may contribute to the difficulties some autistic children have with motor impairment, and that this may be worthy of further research.
Based on the evidence suggested above, I believe that there are reasonable grounds for suggesting that the major neurological and behavioural issues related to ASD may be the result of developmental consequences linked to ACC and AI dysfunction, with follow-on consequences related to ACC-frontopolar cortex connectivity and development. Further, these issues may be linked to a dysfunction in VEN development. VEN issues are not ‘required’ to account for the ACC and AI dysfunction suggested above, but I would suggest – assuming that there is some validity in this analysis – that it may be more than circumstantial that these three ‘final common pathways’ are all linked through either the presence of VENs or potentially via VEN connectivity (assuming ACC – frontopolar cortex links as per Allman et al (2002)). If VEN issues do exist then they may be a result rather than the primary cause of ACC, AI and frontopolar dysfunction, e.g. prior issues may impact VEN migration and development. Regardless, given the potentially significant role of VENs human neurology, any VEN disruption may have significant follow-on consequences.
I would suggest, based on the above, that the ACC, AI, frontopolar cortex and VENs are worthy of significantly more research by those seeking to understand the causes and consequences of ASD.
Further, if this analysis has some validity then I would suggest that it may require some rethinking about the nature of ASD. For a start, it would link ASD’s current DSM-IV behavioural criteria to a neurological foundation and a potential chain of developmental events. It would also call into question some of the more pejorative historical assumptions about ASD and those who qualify for the diagnosis. I will explore these issues in a follow-on post related to ‘Acceptance’.
Allman JM, Hakeem A, Erwin JM, Nimchinsky E, Hof P. (2001). "The anterior cingulate cortex: the evolution of an interface between emotion and cognition", Ann N Y Acad Sci 935:107–17. (Link)
Allman JM, Hakeem A, Watson K (2002) "Two phylogenetic specializations in the human brain." Neuroscientist 8, 335–345 (Link)
Allman JM, Watson KK, Tetreault NA, Hakeem AY (2005) "Intuition and autism: a possible role for von Economo neurons." Trends Cogn Sci 9(8):367–373. (Link)
Ashwin C, Baron-Cohen S, Wheelwright S, O’Riordan M, Bullmore ET (2007) "Differential activation of the amygdala and the 'social brain' during fearful face-processing in Asperger Syndrome" Neuropsychologia (2007) Jan 7;45(1):2-14 (Link)
Back E, Ropar D, Mitchell P (2007) "Do the eyes have it? Inferring mental States from animated faces in autism." Child Dev 2007 Mar-Apr;78(2):397-411. (Link)
Bechara A (2002) "The neurology of social cognition" Brain .2002 Aug:125(Pt8):1673-5. (Link)
Belmonte, M and Carper, R, “Neuroanatomical and Neurophysiological Clues to the Nature of Autism”, Neuroimaging in Child Neuropsychicatric Disorders, Springer-Verlag, April 1998. (Link)
Bielsky IF, Hu SB, Szegda KL, Westphal H, Young LJ (2004) "Profound Impairment in Social Recognition and Reduction in Anxiety-Like Behavior in Vasopressin V1a Receptor Knockout Mice" Neuropsychopharmacology 2004 Mar;29(3):483-93 (Link)
Bischoff-Grethe et al (2002) "Cerebellar Involvement in Response Reassignment Rather than Attention", The Journal of Neuroscience, January 15, 2002, 22(2):546-553 (Link)
Blair RJ (2003) "Facial expressions, their communicatory functions and neuro-cognitive substrates" Philos Trans R Soc Lond B Biol Sci. 2003 Mar 29;358(1431):561-72 (Link)
Blair RJ, Frith U, Smith N, Abell F, Cipolotti L (2002) "Fractionation of visual memory: agency detection and its impairment in autism" Neuropsychologia 2002;40(1):108-18. (Link)
Carlson SM, Moses LJ (2001) "Individual differences in inhibitory control and children's theory of mind." Child Dev 2001 Jul-Aug;72(4):1032-53. (Link)
Carpenter M, Tomasello M, Striano T (2005) "Role Reversal Imitation and Language in Typically Developing Infants and Children with Autism" Infancy, 8(3), 253-278 (Link)
Carpenter M, Pennington BF, Rogers SJ (2002) "Interrelations among social-cognitive skills in young children with autism" J Autism Dev Disord. 2002 Apr;32(2):91-106. (Link)
Carpenter M, Pennington BF, Rogers SJ (2001) "Understanding of others' intentions in children with autism." J Autism Dev Disord. 2001 Dec;31(6):589-99. (Link)
Carr L, Iacoboni M, Dubeau MC, Mazziotta JC, Lenzi GL.(2003) "Neural mechanisms of empathy in humans: a relay from neural systems for imitation to limbic areas." Proc Natl Acad Sci U S A. 2003 Apr 29;100(9):5497-502. Epub 2003 Apr 7 (Link)
Casanova M (2006) “Neuropathological and genetic findings in autism: the significance of a putative minicolumnopathy” Neuroscientist. 2006 Oct; 12(5): 435-41
Chiung-Chih Chang, MD; Yu Chang Lee, MD; Chun-Chung Lui, MD; Shung-Lon Lai, MD, PhD (2007) "Right Anterior Cingulate Cortex Infarction and Transient Speech Aspontaneity" in Arch Neurol. 64:442-446 (Link)
Craig, A.D. (2004) "Human feelings: why are some more aware than others?" Trends Cogn. Sci. 8, 239–241 (Link)
Dalton KM, Nacewicz BM, Johnstone T, Schaefer HS, Gernsbacher MA, Goldsmith HH, Alexander AL, Davidson RJ (2005) "Gaze fixation and the neural circuitry of face processing in autism" Nat Neurosci. 2005 Apr;8(4):519-26. Epub 2005 Mar 6 (Link)
Decety J, Jackson P (2004) "The Functional Architecture of Human Empathy" Behavioral and Cognitive Neuroscience Reviews Vol 3 No 2, June 2004 71-100. (Link)
Devue C, Collette F, Balteau E, Dequeldre C, Luxen A, Maquet P, Bredart S (2007) "Here I Am: The cortical correlates of visual self-recognition" Brain Res 2007 Apr 27;1143:169-82 (Link)
Devinsky O, Morrel M.J., Vogt B.A. (1995) "Contributions of anterior cingulate cortex to behaviour" Brain (1995), 118, 279-306 (Link)
Eslinger PJ, Flaherty-Craig CV, Benton AL (2004) "Developmental outcomes after early prefrontal cortex damage." Brain Cogn 2004 Jun;55(1):84-103. (Link)
Farrer C, Franck N, Georgieff N, Frith CD, Decety J, Jeannerod M (2003) "Modulating the experience of agency: a positron emission tomography study" Neuroimage. 2003 Feb;18(2):324-33 (Link)
Farrer C, Frith CD (2002) "Experiencing oneself vs another person as being the cause of an action: the neural correlates of the experience of agency." Neuroimage 2002 Mar;15(3):596-603 (Link)
Frith U, Blakemore SJ (2004?) "Social Cognition" Foresight Cognitive Systems Project Research Review (Link)
Frith U, Frith CD (2003) "Development and neurophysiology of mentalising" Phil Trans R. Soc. London, B (2003) 358, 459-473 (Link)
Frith U (2001) "Mind blindness and the brain in autism" Neuron. 2001 Dec 20;32(6):969-79 (Link)
Gallese V, Keysers C, Rizzolatti G (2004) "A unifying view of the basis of social cognition" Trends Cogn Sci. 2004 Sep;8(9):396-403 (Link)
Gernsbacher MA (2006) "Toward a Behavior of Reciprocity" (Link)
Gernsbacher MA, Frymiare JL (2001) "Does the Autistic Brain Lack Core Modules" (Link)
Greene JD, Sommerville RB, Nystrom LE, Darley JM, Cohen JD (2001) "An fMRI investigation of emotional engagement in moral judgement." Science 293:2105–8. (Link)
Hadjikhani N, Joseph RM, Snyder J, Tager-Flusberg H (2007) "Abnormal activation of the social brain during face perception in autism" Hum Brain Mapp. 2007 May;28(5):441-9. (Link)
Hadjikhani N, Joseph RM, Snyder J, Chabris CF, Clark J, Steele S, McGrath L, Vangel M, Aharon I, Feczko E, Harris GJ, Tager-Flusberg H. (2004) "Activation of the fusiform gyrus when individuals with autism spectrum disorder view faces" Neuroimage. 2004 Jul;22(3):1141-50. (Link)
Hamilton, A. F. de. C (2007 in press) "Emulation and mimicry for social interaction: A theoretical approach to imitation in autism" Quarterly Journal of Experimental Psychology (Link)
Hamilton A F de C, Brindley RM, Frith U (2007 in press) "Imitation and action understanding in autistic spectrum disorders: How valid is the hypothesis of a deficit in the mirror neuron system?" Neuropsychologia (2007) doi:10.1016/j.neuropsychologia.2006.11.022 (Link)
Hill E (2004a) "Executive dysfunction in autism" Trends Cogn. Sci. 8, 26–32 (Link)
Hill E (2004b) "Evaluating the theory of executive dysfunction in autism" Developmental Review, 24; 189-233 (Link)
Hobson JA, Hobson RP (2007) "Identification: The missing link between joint attention and imitation?" Dev Psychopathol. 2007 Spring;19(2):411-31 (Link)
Hoffman EA, Haxby JV (2000) "Distinct representations of eye gaze and identity in the distributed human neural system for face perception" Nat Neurosci (2000) Jan;3(1):80-4 (Link)
Jackson P, Brunet E, Meltzoff AN, Decety J (2006) "Empathy examined through the neural mechanisms involved in imagining how I feel versus how you feel pain" Neuropsychologia 44 (2006) 752-761 (Link)
Kawashima R, Sugiura M, Kato T, Nakamura A, Hatano K, Ito K, Fukuda H, Kojima S, Nakamura K. (1999) "The human amygdala plays an important role in gaze monitoring. A PET study" Brain. 1999 Apr;122 ( Pt 4):779-83 (Link)
Kelly AM, Hester R, Murphy K, Javitt DC, Foxe JJ, Garavan H (2004) "Prefrontal-subcortical dissociations underlying inhibitory control revealed by event-related fMRI" Eur J Neurosci 2004 Jun;19(11):3105-12 (Link)
Kennedy DP, Semendeferi K, Courchesne E (2007) "No reduction of spindle neuron number in frontoinsular cortex in autism" Brain and Cognition, doi:10.1016/j.bandc.2007.01.007
Kern JK, Trivedi MH, Grannemann BD, Garver CR, Johnson DG, Andrews AA, Savla JS, Mehta JA, Schroeder JL. (2007) "Sensory correlations in autism" Autism. 2007 Mar;11(2):123-34. (Link)
Kilner JM, Marchant JL, Frith CD (2006) “Modulation of the mirror system by social relevance” Soc Cogn Affect Neurosci. 2006;1(2):143-148. (Link)
Kylliainen A, Hietanen JK.(2006) "Skin conductance responses to another person's gaze in children with autism" J Autism Dev Disord. 2006 May;36(4):517-25 (link)
Ming X, Brimacombe M, Wagner GC (2007) "Prevalence of motor impairment in autism spectrum disorders" Brain Dev. 2007 Apr 27; [Epub ahead of print] (Link)
Nimchinsky E, Gilissen E, Allman JM, Perl DP, Erwin JM, Hof PR. (1999). "A neuronal morphologic type unique to humans and great apes", Proc Nat Acad Sci 96:5268–73. (Link)
Pelphrey KA, Morris JP, McCarthy G. (2005) "Neural basis of eye gaze processing deficits in autism" Brain. 2005 May;128(Pt 5):1038-48. Epub 2005 Mar 9 (Link)
Pierce K, Haist F, Sedaghat F, Courchesne E.(2004) "The brain response to personally familiar faces in autism: findings of fusiform activity and beyond", Brain. 2004 Dec;127(Pt 12):2703-16. Epub 2004 Aug 19 (Link)
Posner M.I., Rothbart M.K., "Attention, self-regulation and consciousness" Phil.Trans. R. Soc. Lond. B (1998) 353, 1915-1927 (Link)
Ravizza SM, Ivry R.B. (2001) "Comparison of the basal ganglia and cerebellum in shifting attention" in J Cogn Neurosci, 2001 Apr 1;13(3):285-97. (Link)
Rutgers AH, Bakermans-Kranenburg MJ, van Ijzendoorn MH, van Berckelaer-Onnes IA (2004) "Autism and attachment: a meta-analytic review" J Child Psychol Psychiatry 2004 Sep;45(6):1123-34. (Link)
Sasson NJ (2006) "The development of face processing in autism." J Autism Dev Disord. 2006 Apr;36(3):381-94 (Link)
Sasson N, Tsuchiya N, Hurley R, Couture SM, Penn DL, Adolphs R, Piven J.(2007) "Orienting to social stimuli differentiates social cognitive impairment in autism and schizophrenia." Neuropsychologia 2007 Mar 15; (in press) (Link)
Schore, von Allan N (2001) "The Effects of Early Relational Trauma on Right Brain Development, Affect Regulation, and Infant Mental Health" AGSP Jahrgang 2001 (Link)
Schultz RT, Grelotti DJ, Klin A, Kleinman J, Van der Gaag C, Marois R, Skudlarski P (2003) "The role of the fusiform face area in social cognition: implications for the pathobiology of autism" Philos Trans R Soc Lond B Biol Sci. 2003 Feb 28;358(1430):415-27 (Link)
Schumann C.M., Amaral D.G. (2006) "Stereological Analysis of Amygdala Neuron Number in Autism" J Neurosci 26(29):7674 –7679 (Link)
Singer T, Seymour B, O’Doherty J, Kaube H, Dolan RJ, Frith CD (2004) "Empathy for pain involves the affective but not sensory components of pain." Science 303, 1157–1162 (Link)
Spezio ML, Huang PY, Castelli F, Adolphs R. (2007) "Amygdala damage impairs eye contact during conversations with real people" J Neurosci. 2007 Apr 11;27(15):3994-7 (Link)
Strange B.A., Henson R.N.A., Friston K.J., Dolan R,J. (2001), "Anterior Prefrontal Cortex Mediates Rule Learning in Humans" Cerebral Cortex Nov 2001;11:1040-1046 (Link)
Stieben J, Lewis MD, Granic I, Zelazo PD, Segalowitz S, Pepler D (2007) "Neurophysiological mechanisms of emotion regulation for subtypes of externalizing children" Dev Psychopathol 2007 Spring;19(2):455-80. (Link)
Watson K.K.; Jones T.K., Allman J.M. (2006) "Dendritic Architecture Of The Von Economo Neurons" Neuroscience 141 1107–1112 (Link)
Wicker B, Keysers C, Plailly J, Royet JP, Gallese V, Rizzolatti G (2003) "Both of us disgusted in My insula: the common neural basis of seeing and feeling disgust" Neuron, 40, 655-664. (Link)
Williams D.L, Goldstein G., Minshew N.J. (2006) "Neuropsychologic Functioning In Children With Autism: Further Evidence For Disordered Complex Information-Processing", Child Neuropsychol. 2006 August ; 12(4-5): 279–298. (Link)
Williams JH, Waiter GD, Perra O, Perrett DI, Whitten A (2005) "An fMRI study of joint attention experience" Neuroimage 2005 Mar; 25(1):133-40 (Link)
Williams MA, McGlone F, Abbott DF, Mattingley JB (2005) "Differential amygdala responses to happy and fearful facial expressions depend on selective attention" Neuroimage. 2005 Jan 15;24(2):417-25 (Link)