The amygdala, a small almond-shaped structure deep in the brain’s temporal lobe, is one of the most consistently implicated brain regions in autism research. Decades of imaging and postmortem studies have shown that the amygdala develops along a different trajectory in autistic individuals, growing too large too early in life and then failing to keep pace with typical growth later on. But what looked at first like a straightforward story of one brain region going awry has turned out to be far more complex, involving shifts in neuron count, altered wiring to other brain areas, and connections to everything from face perception to sensory overload to anxiety.
An Unusual Growth Curve
One of the most replicated findings in autism neuroscience is that the amygdala is enlarged in young autistic children but not in autistic adolescents or adults. A study comparing children and adolescents found that autistic children between roughly 7.5 and 12.5 years old had larger amygdala volumes on both sides of the brain compared to non-autistic children of the same age, but by adolescence that size difference had disappeared. The reason was not that the autistic amygdala shrank; rather, the typically developing amygdala undergoes substantial growth during this period while the autistic amygdala does not, so the two groups converge.1PubMed Central. The amygdala is enlarged in children but not adolescents with autism; the hippocampus is enlarged at all ages
More recent work pushed the timeline back even further. Researchers studying infants at high familial likelihood for autism found that those who went on to receive a diagnosis had typically sized amygdalas at six months of age. Between six and 24 months, however, amygdala growth accelerated dramatically compared to all other groups. By 12 months, autistic infants already had measurably larger amygdalas, and the rate of growth in that six-to-twelve-month window predicted the severity of social difficulties at age two.2PubMed Central. Subcortical Brain Development in Autism and Fragile X Syndrome: Evidence for Dynamic, Age- and Disorder-Specific Trajectories in Infancy This finding is striking because it suggests amygdala overgrowth is among the earliest detectable brain differences in autism, appearing before most behavioral signs are clinically recognized.
What Is Happening Inside the Cells
Volume is just the beginning. When researchers have examined postmortem amygdala tissue, the cellular picture gets more nuanced. In typically developing brains, neuron numbers in the amygdala increase from birth through adulthood, likely through the continued maturation and migration of neurons from a region called the paralaminar nucleus. In autistic individuals, however, there appears to be an initial excess of neurons during childhood followed by a reduction in adulthood across multiple amygdala subregions.3PubMed Central. Neuron numbers increase in the human amygdala from birth to adulthood, but not in autism This mirrors the volume findings: too much, too soon, followed by a plateau or decline.
Stereological analyses, which count neurons in precisely defined tissue samples, have confirmed that adult autistic amygdalas contain fewer neurons overall and particularly in the lateral nucleus, one of the amygdala’s main input zones for sensory and social information.4PubMed Central. Stereological analysis of amygdala neuron number in autism So the childhood enlargement does not translate into a lasting advantage in cell count. If anything, the early overgrowth may disrupt the normal pruning and maturation processes that keep the region functioning smoothly.
There is also a dendritic story worth knowing. Dendrites are the branching extensions through which neurons receive signals, and spines are the tiny protrusions on dendrites where most of those signals arrive. In young autistic individuals under 18, spine density on amygdala neurons is higher than in age-matched controls. But as autistic individuals age into adulthood, spine density drops until it is indistinguishable from that of non-autistic adults. Typically developing brains, by contrast, do not show this decline.5PubMed Central. Protracted dendritic growth in the typically developing human amygdala and increased spine density in young ASD brains This excess of connection points during childhood could contribute to the amygdala being overly reactive or poorly tuned during a critical window for social learning.
Wiring That Runs Too Hot or Too Cold
The amygdala does not operate in isolation. It sends and receives signals from dozens of other brain regions, and those communication pathways are consistently altered in autism. Importantly, the pattern is not simply “more” or “less” connectivity across the board. It depends on which part of the amygdala you look at.
The amygdala has several subregions, each with different roles. Broadly, the laterobasal portion receives incoming sensory and social information, while the centromedial and superficial portions handle output, including autonomic responses like changes in heart rate and emotional reactions. Resting-state brain imaging studies have found that in autistic individuals, the laterobasal subregion shows decreased connectivity with other brain areas, while the centromedial and superficial subregions show increased connectivity.6PubMed Central. Subregional differences in intrinsic amygdala hyper and hypo connectivity in autism spectrum disorder In plain terms, the input side of the amygdala is underconnected while the output side is overconnected. One way to think about this: the amygdala may be generating strong reactions without having adequately processed the incoming information that should be guiding those reactions.
A multi-site imaging study using dynamic (time-varying) connectivity measures painted a similar picture. The basolateral amygdala in autistic participants showed weaker-than-typical connections to temporal and frontal regions involved in language and higher-order cognition, while certain other connections were stronger than typical.7PubMed. Altered resting state dynamic functional connectivity of amygdala subregions in patients with autism spectrum disorder: A multi-site fMRI study This fits a broader pattern in autism research where long-range connections between distant brain regions tend to be weaker while some local or short-range connections are preserved or even stronger.
One of the most studied amygdala pathways involves the medial prefrontal cortex, a region that helps regulate emotions and guide social behavior. In autistic children, the signal from the prefrontal cortex to the amygdala is weaker than in non-autistic children, and the degree of that weakening tracks with the severity of social difficulties.8PubMed. Attenuated link between the medial prefrontal cortex and the amygdala in children with autism spectrum disorder: Evidence from effective connectivity within the “social brain” Think of the prefrontal cortex as a manager that helps the amygdala calibrate its responses. With a weaker line of communication, the amygdala’s responses to social signals are less modulated.
The amygdala’s link to the anterior insula, a region involved in detecting what matters in the moment, is also reduced in autism. Resting-state analyses have shown weaker amygdala-to-insula connectivity in autistic individuals compared to controls, a finding that aligns with theories about disrupted salience processing.9Social Cognitive and Affective Neuroscience. Reduced functional connectivity within and between ‘social’ resting state networks in autism spectrum conditions
Face Perception and the Gaze Problem
A longstanding observation in autism research is that many autistic individuals process faces differently. The amygdala plays a central role in evaluating faces, especially for emotional content, and imaging studies have revealed a telling quirk: autistic individuals often show elevated amygdala activation not just to emotional faces but to neutral ones. In one study, the source of a significant group-by-emotion interaction in amygdala activity was the unusually high response to neutral faces in the autistic group.10Social Cognitive and Affective Neuroscience. Elevated amygdala response to faces and gaze aversion in autism spectrum disorder
This matters because neutral faces are ambiguous. When the amygdala responds strongly to an ambiguous social signal, it can make that signal feel threatening or at least effortful to process. Some researchers believe this heightened amygdala reactivity to faces contributes to gaze avoidance, a common (though far from universal) feature of autism. Looking away from someone’s eyes may be a regulatory strategy to reduce the amygdala’s stress-like response, not a sign of disinterest. This reframing has practical implications: therapies that pressure eye contact without addressing the underlying discomfort may inadvertently increase distress rather than improve social engagement.
Anxiety and Social Difficulties Pull in Opposite Directions
One of the more surprising findings in this field is that anxiety and core social difficulties in autism have independent and opposing relationships with amygdala activity. In autistic individuals, higher anxiety was associated with greater amygdala activation, while greater social deficits were associated with lower amygdala activation.11Social Cognitive and Affective Neuroscience. Anxiety and social deficits have distinct relationships with amygdala function in autism spectrum disorder
This finding has real implications for how we think about the amygdala’s role. If the amygdala were simply “broken” in autism, you might expect all autism-related difficulties to correlate with amygdala activity in the same direction. Instead, the picture is more like two partially independent systems operating through the same structure. Anxiety, which involves hypervigilance and threat detection, revs the amygdala up. Social disengagement, which involves reduced attention to social cues, is linked to a quieter amygdala. For clinicians, this means that an autistic person with significant anxiety and an autistic person with primarily social communication differences may have very different amygdala profiles, even though both carry the same diagnosis.
Sensory Overload and the Amygdala
Sensory overresponsivity, the experience of everyday sights, sounds, or textures as painfully intense, affects a large proportion of autistic individuals. Brain imaging in autistic youth found that the severity of sensory overresponsivity correlated with greater activation in the right amygdala as well as somatosensory and parietal areas.12PubMed Central. Neurobiology of Sensory Overresponsivity in Youth With Autism Spectrum Disorders This suggests the amygdala is not only involved in processing social information but also contributes to the distress response triggered by non-social sensory input. If you have ever seen an autistic child cover their ears in a noisy restaurant and wondered whether that reaction was “emotional” or “sensory,” the answer, at least neurologically, is both. The amygdala sits at the intersection of sensory and emotional processing, and its altered function in autism appears to amplify both.
Sex Differences in Amygdala Patterns
Autism is diagnosed roughly three to four times more often in males than females, and there is growing evidence that the amygdala contributes differently to autistic traits depending on biological sex. A study of amygdala resting-state connectivity in children found that the differences between autistic and non-autistic children were greater in girls than in boys. Additionally, autistic children showed reduced sex-typical patterns in amygdala connectivity, meaning the usual male-female differences in how the amygdala connects to other regions were blunted in the autistic group.13PubMed Central. Sex Differences in the Amygdala Resting State Connectome of Children with Autism Spectrum Disorder
Longitudinal work tells a complementary story. A study tracking brain development over time found that autistic males and females showed different patterns of altered development in regions connected to the amygdala. In males, the subgenual anterior cingulate cortex (a region involved in mood regulation) was most affected; in females, the fusiform gyrus and superior temporal gyrus (regions involved in face and voice processing) showed the largest differences.14PubMed Central. Altered Development of Amygdala-Connected Brain Regions in Males and Females with Autism These differences could help explain why autistic females often present with a different clinical profile than autistic males, with some research showing girls are more likely to have co-occurring internalizing symptoms like anxiety and depression.
On that point, a study of young autistic children found that a higher proportion of girls (about 40%) fell into a high-psychopathology subgroup compared to boys (about 22%), and amygdala enlargement was specifically associated with that subgroup. Furthermore, right amygdala volume was positively associated with internalizing behaviors in girls but not in boys.15PubMed Central. High Psychopathology Subgroup in Young Children With Autism: Associations With Biological Sex and Amygdala Volume This is a reminder that amygdala differences in autism are not monolithic. They interact with sex, symptom profile, and age in ways that a one-size-fits-all model cannot capture.
The Limits of an Amygdala-Centered Theory
The “amygdala theory of autism,” first proposed in the early 2000s, suggested that amygdala dysfunction was the core neural basis for the social difficulties seen in autism. It was an appealing idea, given the amygdala’s well-established role in processing emotional and social information. But two decades of data have made it clear the theory is incomplete.
A 2023 review evaluating the theory’s track record found that the amygdala could account for a limited subset of autistic traits, primarily those related to face perception, but could not explain broader patterns like differences in social attention. The authors argued that a network-level view is more appropriate, one that considers how the amygdala interacts with the prefrontal cortex, insula, temporal regions, and other areas across the whole brain.16PubMed Central. A revisit of the amygdala theory of autism: Twenty years after
An alternative framework, the “relevance detector” theory, proposes that the amygdala is not specifically a social-emotion processor but rather part of a circuit that identifies what matters to the individual in any given moment. Under this model, the behavioral features of autism reflect disruption in a broader relevance-detection network, not just a social one, which would better explain why autistic traits extend beyond social difficulties into areas like restricted interests, repetitive behaviors, and sensory sensitivities.17PubMed Central. The amygdala and the relevance detection theory of autism: an evolutionary perspective Neither theory captures the full picture, but the shift from “one region is broken” to “a whole network develops differently” is probably the most important conceptual update in this area over the past two decades.
Molecular and Chemical Layers
Beneath the structural and connectivity findings, researchers have begun probing what is happening at the molecular level. Gene expression analyses of postmortem amygdala tissue have identified clusters of genes with altered expression in the autistic amygdala, especially within excitatory neurons.18PubMed Central. Identification of amygdala-expressed genes associated with autism spectrum disorder This fits with a broader hypothesis in autism research that the balance between excitatory and inhibitory signaling in certain brain circuits is disrupted. Animal model work using rats exposed to valproic acid, which produces some autism-like behaviors, has shown a shifted excitatory-to-inhibitory balance specifically in the lateral amygdala.19PLoS ONE. The Amygdala Excitatory/Inhibitory Balance in a Valproate-Induced Rat Autism Model
Oxytocin, sometimes called the “bonding hormone,” also enters the picture through the amygdala. In autistic adults, the connectivity between the amygdala and the hippocampus has been linked to endogenous oxytocin levels, and administering oxytocin can alter this pathway.20PubMed. Amygdala-Hippocampal Connectivity Is Associated With Endogenous Levels of Oxytocin and Can Be Altered by Exogenously Administered Oxytocin in Adults With Autism This does not mean oxytocin is a treatment for autism, and clinical trials of intranasal oxytocin have had mixed results overall. But the finding illustrates that the amygdala’s connectivity patterns in autism are not fixed. They respond to neurochemical signals, which leaves room for interventions even if none have proven broadly effective yet.
Early Steps Toward Targeted Interventions
One small but intriguing intervention study tested bumetanide, a diuretic that also modifies the chloride balance in neurons, essentially nudging the excitatory-inhibitory ratio in a direction that could quiet overactive circuits. After treatment, autistic participants showed a significant decrease in amygdala activation during tasks requiring eye contact compared to their pre-treatment baseline. The effect was specific: it showed up only during constrained eye-gaze conditions, not during free-viewing conditions or in brain regions unrelated to face processing.21Scientific Reports. Bumetanide for autism: more eye contact, less amygdala activation The sample was small, and bumetanide is not approved for autism, but the study illustrates a principle: if the amygdala’s overreactivity during social processing can be dialed down pharmacologically, it could ease the neurological cost of social interaction for some autistic individuals without requiring them to simply power through discomfort.
Whether future interventions target the amygdala directly, its connections to the prefrontal cortex, or the neurochemical environment that shapes its function, the research consistently points toward the same underlying issue. The amygdala in autism is not simply too big or too active. It develops on a different schedule, wires itself differently to the rest of the brain, and responds to the world in ways that make some experiences, particularly social ones, more effortful and sometimes more distressing. Understanding that trajectory in detail is what makes the difference between a blunt “fix the broken part” approach and something more nuanced that works with the autistic brain rather than against it.