A Neurofeedback therapy of facial expression recognition in Autism shifts connectivity to higher levels within the third visual pathway in relation to clinical improvements

A five-session neurofeedback therapy targeting the posterior superior temporal sulcus in individuals with autism spectrum disorder induces clinical improvements by reorganizing functional connectivity within the third visual pathway, specifically decreasing links to low-level visual areas while strengthening connections with higher-level social cognition regions.

The Gist

The Big Picture: Rewiring the Brain's "Social GPS"

Imagine the human brain is a massive, bustling city with thousands of roads connecting different neighborhoods. In people with Autism Spectrum Disorder (ASD), some of the main roads used for understanding social cues—like reading a friend's face or recognizing a smile—are often clogged, broken, or taking the wrong exit.

This study tested a new kind of "traffic control" called Neurofeedback. The goal was to see if we could teach the brain to reroute its traffic, specifically focusing on a neighborhood called the pSTS (posterior Superior Temporal Sulcus). Think of the pSTS as the city's "Social Hub," a place dedicated to understanding emotions and facial expressions.

The Problem: A Traffic Jam in the "Low-Level" Districts

Before the treatment, the researchers noticed that the Social Hub (pSTS) was getting too much traffic from the "Low-Level Visual Districts" (areas like the OFA that just see basic shapes and dots) and not enough traffic from the "High-Level Social Districts" (areas that understand complex feelings and context).

It was like a restaurant kitchen where the chef (the Social Hub) was spending all their time arguing with the dishwasher about the shape of the plates, instead of focusing on cooking the meal (understanding the emotion).

The Solution: The "Brain Gym" (Neurofeedback)

The researchers put 15 young men with autism into a 5-session "Brain Gym."

• The Setup: They used an MRI machine to watch the brain in real-time.
• The Task: The participants had to imagine different facial expressions (happy, sad, neutral).
• The Feedback: As they imagined these faces, a computer avatar on a screen would change its expression based on how well they were activating their Social Hub (pSTS). If they did it right, the avatar smiled; if not, it looked confused.

It's like playing a video game where your brain is the controller. The goal was to learn how to "play" the game by consciously turning up the volume on the Social Hub.

The Results: A Major Shift in Traffic Flow

After just five sessions, the researchers scanned the brains again and found a massive reorganization of the city's roads:

1. Clearing the Low-Level Jam: The connection between the Social Hub and the "Low-Level Visual Districts" (the basic shape detectors) actually decreased. The brain stopped wasting energy arguing about basic shapes.
2. Building High-Speed Highways: The connection between the Social Hub and the "High-Level Social Districts" (like the FFA and mSTS, which handle complex emotions and social context) increased.
3. The "Third Visual Pathway": The study highlights a specific route in the brain called the "Third Visual Pathway," which is specialized for social information. The treatment successfully shifted the brain's focus onto this highway, moving away from the old, clogged routes.

The Analogy: Imagine the brain was previously trying to read a book by staring at the individual letters (low-level) and getting confused. After the training, the brain learned to look at the whole sentences and paragraphs (high-level), making the story (social interaction) much easier to understand.

Did It Help? Yes!

The most exciting part is that these brain changes weren't just random; they were directly linked to real-world improvements.

• The Scorecard: The participants' caregivers filled out a checklist (ATEC) measuring autism symptoms.
• The Link: The participants whose brains showed the biggest "traffic shift" (more high-level connections, fewer low-level ones) were the ones who showed the biggest improvements in their daily lives and ability to recognize fear or happiness in faces.

The Takeaway

This study suggests that the brain is like a flexible muscle. Even in autism, where social processing can feel like navigating a maze, we can use Neurofeedback to teach the brain new routes. By training a specific "Social Hub," we didn't just fix that one spot; we reorganized the entire network, helping the brain move from getting stuck on details to understanding the bigger social picture.

It's a hopeful sign that with the right "traffic training," the brain can learn to drive itself toward better social connection.

Technical Summary

1. Problem Statement

Autism Spectrum Disorder (ASD) is characterized by significant deficits in social communication, particularly in the recognition of facial expressions. While the "third visual pathway" (a social perception stream involving the superior temporal sulcus, STS) is hypothesized to be critical for processing dynamic social cues, its functional integrity in ASD remains poorly understood. Previous research suggests underconnectivity between key regions like the posterior STS (pSTS) and middle STS (mSTS) in ASD.

The study addresses a critical gap: Does a targeted neurofeedback (NF) intervention aimed at the pSTS induce specific, adaptive reorganization of functional connectivity (FC) within the face-processing network, and are these neural changes correlated with clinical improvements in social behavior?

2. Methodology

Participants and Design

• Subjects: 15 male participants with ASD (mean age ~19 years).
• Intervention: A 5-session real-time fMRI neurofeedback (NF) clinical trial (NCT02440451).
• Target Region: The right posterior Superior Temporal Sulcus (pSTS), identified as a core hub for the third visual pathway and emotional facial expression processing.
• Task: Participants performed an imagery task (imagining happy, sad, and neutral expressions) while receiving real-time visual feedback (an avatar morphing) based on their pSTS activity levels.

Data Acquisition & Preprocessing

• Scanner: 3T Siemens Magnetom Tim Trio.
• Protocol: Included a functional localizer (to define individual pSTS ROIs) and four imagery runs per session (training, two NF runs, one transfer run without feedback).
• Preprocessing: Performed using CONN and SPM. Steps included realignment, slice-timing correction, outlier removal (ART), normalization to MNI space, segmentation, and smoothing (8mm FWHM). Confounds (white matter, CSF, motion) were regressed out.

Analysis Strategy

The study employed two primary connectivity analysis approaches comparing Session 1 (Baseline) vs. Session 5 (Post-treatment):

1. ROI-to-ROI Analysis: Examined the "extended face network" (14 regions defined by Fox et al., 2009, including FFA, OFA, pSTS, mSTS, IFG, Amygdala, etc.).
2. Seed-to-Voxel Analysis: Used the individualized pSTS as the seed to map whole-brain connectivity changes during both the localizer (passive viewing) and transfer runs (imagery without feedback).
3. Behavioral Correlation: Correlated changes in functional connectivity with changes in clinical scores:
   • ATEC (Autism Treatment Evaluation Checklist): Lower scores indicate improvement.
   • FEEST: Facial Expressions of Emotion test (specifically fear recognition).

3. Key Contributions

• Mechanistic Insight into NF: Demonstrates that NF targeting a single node (pSTS) induces widespread network reconfiguration, not just local changes.
• Validation of the Third Visual Pathway: Provides empirical evidence that ASD involves a disruption in the social visual pathway (pSTS-mSTS) and that therapeutic intervention can restore connectivity within this specific hierarchy.
• Connectivity-Behavior Link: Establishes a direct statistical link between specific FC shifts (decoupling from low-level visual areas and strengthening high-level social networks) and clinical symptom reduction.

4. Key Results

A. Network Reorganization (Localizer & Transfer Runs)

• Decreased Low-Level Connectivity: Significant reduction in connectivity between the pSTS and low-level visual areas, specifically the Occipital Face Area (OFA) and Lateral Occipital Cortex. This suggests a "dissociation" from early visual processing stages.
• Increased High-Level Connectivity: Significant increase in connectivity between the pSTS and higher-order regions:
  • Within the Third Pathway: Increased connection between the right pSTS and the right middle STS (mSTS), and between the right mSTS and the left FFA.
  • Social Cognition Networks: Increased connectivity with the Temporoparietal Junction (TPJ), including the Supramarginal Gyrus and Angular Gyrus.
  • Memory/Emotion: Increased connectivity with the Parahippocampal Gyrus.
• Graph Theory: An increase in the clustering coefficient of the right pSTS was observed, indicating enhanced local integration within the network.

B. Parametric Trends

• Analysis across all 5 sessions revealed a progressive decrease in connectivity between the pSTS and ventral/inferotemporal visual areas over time, reinforcing the shift away from low-level processing.

C. Correlation with Clinical Outcomes (ATEC)

• Negative Correlation (Improvement): Greater clinical improvement (lower ATEC scores) was associated with increased connectivity between the pSTS and regions involved in social cognition and Theory of Mind (ToM), specifically the Supramarginal Gyrus, Superior Temporal Sulcus, and Angular Gyrus (right lateralized).
• Positive Correlation (Improvement): Improvement was also linked to increased connectivity with the Precuneus and Posterior Cingulate (parts of the Default Mode Network), suggesting a treatment-induced decoupling or reorganization of DMN interactions with the social network.
• Imagery Task Specifics: In the transfer run (imagery without feedback), clinical improvement correlated with increased connectivity between the pSTS and occipital visual areas (Lingual Gyrus, Occipital Pole), suggesting that successful NF training optimizes the recruitment of the "social visual pathway" even during internal imagery.

5. Significance and Conclusion

This study provides robust evidence that neurofeedback targeting the pSTS can effectively rewire the functional architecture of the brain in individuals with ASD. The findings support a model where successful treatment involves:

1. Shifting Processing Hierarchy: Moving processing load from low-level visual areas (OFA) to higher-level social integration hubs (mSTS, TPJ).
2. Restoring the Third Visual Pathway: Reinforcing the specific pSTS-mSTS connection, which is often underconnected in ASD.
3. Clinical Translation: The observed neural reorganization is not merely a statistical artifact but is directly predictive of behavioral improvements in adaptive functioning.

Conclusion: The study validates the pSTS as a viable therapeutic target for ASD and suggests that network neuroscience approaches can identify biomarkers for treatment response. The "shift" toward higher-level connectivity within the social visual pathway represents a potential mechanism for restoring social cognition deficits. Future multicentric trials are recommended to probe this reorganization as a standard treatment target.