Single-Neuron Correlates of Abnormal Face Processing by the Amygdala In Autism

Background: Face processing impairments in autism are hypothesized to result in part from amygdala dysfunction. During emotional judgments, people with autism fail to use information from the eye region of faces. Similar functional impairments are also apparent in patients with focal amygdala lesions. How neurons in the amygdala respond to the eyes and other features in faces, and how these responses might differ in people with autism, is unknown.

Objectives: To further elucidate the neural basis of face processing abnormalities in people with autism we used small randomly sampled pieces of faces as stimuli. This approach, called “bubbles”, was used by us previously to show that people with autism make abnormal use of the mouth region of faces rather than the eyes. Here we hypothesized that single neurons in the amygdala of people with autism would show a similar abnormality: responses driven more by the mouth than the eyes.

Methods: We recorded 145 single neurons in the amygdala from 6 neurosurgical patients undergoing epilepsy monitoring (2 of whom had a clinical and ADOS-verified diagnosis of autism) during presentation of random small pieces of faces (“bubbles”). Participants were asked to judge as fast as possible whether a presented piece of a face was happy or fearful. We also recorded responses to whole emotional faces, as well as to selected cutouts of eyes and mouth regions. The single-neuron responses were then regressed onto the regions of the face shown in the stimuli to tell us which parts of a face were most potent in eliciting responses from recorded amygdala neurons.

Results:    Behavioral data confirmed that individuals with autism fail to make normal use of information from the eyes within faces, and instead make exaggerated use of the mouth (Spezio, Adolphs, Hurley & Piven (2007), JADD 37:929). A significant fraction of the 145 well-isolated single units responded selectively to full or parts of faces and not to control stimuli of equal contrast/luminance (scrambles). Neuronal classification analysis of spikes from these neurons revealed neurons driven by both the mouth and eye region in participants without a diagnosis of autism. By contrast, participants with autism showed neuronal responses that almost completely lacked eye-region driven responses and were instead driven predominantly by the mouth.

Conclusions: The previously reported bias in autism to overutilize information about the mouth region and underutilize information from the eye region of faces was reflected in the responses of single neurons recorded in the amygdala. This is the first demonstration of a direct link between behavioral judgment of faces and single-neuron responses in the amygdala. The findings provide a neural mechanism that may underlie abnormal face processing in autism.