Summary:
A neurocognitive electroencephalography (EEG) study led by researchers at the Hebrew University of Jerusalem reveals that autistic and non-autistic adults achieve identical perceptual accuracy on auditory tasks, but diverge markedly in how their brains evaluate feedback and update expectations. Published by researchers at the Edmond and Lily Safra Center for Brain Sciences (ELSC), the study identified three distinct neural and behavioral differences: blunted feedback-related negativity (FRN) in response to errors, reduced sensitivity to task difficulty, and weaker weighting of reliable prior trials. Replicated one to two years later, the findings shed light on the computational mechanisms underlying behavioral flexibility in autism.
Key Facts:
- Equal Perceptual Accuracy: Both autistic (n = 33) and non-autistic (n = 35) adults achieved approximately 70% accuracy on fine auditory pitch discrimination tasks, confirming that primary perceptual processing is intact.
- Three Independent Divergences: Autistic participants displayed a significantly reduced feedback-related negativity (FRN) signal following negative feedback, minimal neural modulation based on task difficulty, and significantly less behavioral adjustment following high-confidence correct trials.
- Longitudinal Replicability: Follow-up testing conducted 1 to 2 years later across 51 participants replicated all three neural and behavioral signatures, which when combined in a machine-learning model classified neurotypes with ~80% accuracy.
Source: The Hebrew University of Jerusalem
Navigating rapidly changing social and sensory environments requires the brain to continuously compute prediction errors: it must register mistakes, gauge the confidence of its choices, and determine how heavily past experiences should guide future actions. For autistic individuals, unexpected environmental transitions and shifting contexts can be disorienting and stressful.
While clinical research has historically focused on sensory processing differences or hyper-reactivity, the computational steps linking perception, feedback monitoring, and flexible decision-making have remained poorly understood.
Now, an international investigation led by Prof. Merav Ahissar of the Edmond and Lily Safra Center for Brain Sciences (ELSC) and the Department of Psychology at The Hebrew University of Jerusalem indicates that the primary divergence in autism is not how sensory information is perceived, but how the brain evaluates feedback and weights past confidence.
Equal Pitch Accuracy, Divergent Feedback Circuits
The research team evaluated 33 autistic and 35 non-autistic adult participants on a classical psychoacoustic task. Participants listened to successive pairs of pure auditory tones and identified which tone possessed the higher pitch. Immediately following each response, participants received visual feedback signaling whether their judgment was correct or incorrect, while continuous 64-channel electroencephalography (EEG) monitored event-related potentials (ERPs).
Both cohorts performed with identical competence, averaging roughly 70% task accuracy. This finding demonstrates that basic auditory pitch encoding and discrimination thresholds are equivalent between autistic and non-autistic individuals.
However, electrophysiological recordings during feedback presentation uncovered three distinct neurocognitive divergences:
- Attenuated Feedback-Related Negativity (FRN): When presented with negative feedback indicating a mistake, non-autistic participants generated a pronounced, characteristic negative electrical deflection over frontocentral electrodes. the FRN wave, reflecting anterior cingulate cortex error-processing. In autistic participants, this neural error signal was markedly blunted.
- Insensitivity to Decision Difficulty: In non-autistic individuals, neural feedback potentials shifted dynamically based on whether the preceding auditory judgment had been objectively subtle or obvious. In contrast, the neural response in autistic adults remained largely uniform regardless of trial difficulty.
- Diminished Anchoring on Reliable Success: Following trials that were easy and correct—where non-autistic participants heavily biased their subsequent choices based on trustworthy, high-confidence evidence, autistic participants showed substantially less behavioral reliance on those recent successful outcomes.
“Our findings suggest that the differences associated with autism are not necessarily in the ability to perceive information accurately, but in how the brain evaluates experiences and uses them to guide future decisions,” explained Prof. Ahissar. “We identified several distinct processes that may help us understand why adapting to rapidly changing situations can be more challenging for autistic individuals.”
Stability Across Years and Distinct Cognitive Streams
To verify whether these neural patterns were fleeting state-dependent fluctuations or enduring cognitive traits, the researchers re-tested 51 of the original participants in a follow-up experimental session one to two years later.
All three markers, blunted FRN amplitude, difficulty insensitivity, and reduced reliance on recent successful trials, replicated with high individual stability across both time points.
Statistical modeling revealed that the three measures were not significantly correlated with one another within participants. This indicates that they do not stem from a single overarching cognitive deficit, but rather reflect separate, parallel neurocognitive streams in predictive processing and reward evaluation.
When the investigators integrated these three independent metrics into a supervised machine-learning model, the classifier distinguished between autistic and non-autistic participants with approximately 80% cross-validated accuracy. The authors stress, however, that the model is designed to map neurocognitive variation and is not intended as a stand-alone diagnostic tool.
Rethinking Behavioral Flexibility in Autism
The study provides important empirical support for Bayesian predictive coding theories of autism. Rather than reflecting an inability to learn or an inaccuracy in sensory acquisition, autism may involve a distinct statistical weighting style: the autistic brain treats each discrete event as unique, placing less predictive faith in recent history and blunting the emotional impact of standard performance feedback.
By characterizing how the brain tracks its own decision certainty, the research provides a neurocomputational framework for understanding why routine disruptions and shifting task demands can be taxing for autistic individuals, paving the way for adaptive educational and workplace environments tailored to diverse cognitive profiles.
Editorial Notes:
- This article was edited by a Neuroscience News editor.
- Journal paper reviewed in full.
- Additional context added by our staff.
About this ASD Research:
- Media Contact: Yarden Mills
- Source: Hebrew University of Jerusalem
- Image Credit: Image credited to Neuroscience News
- Original Research is Open Access: Journal of Neuroscience (Oct 26, 2026). “Neurocognitive characteristics of perceptual feedback processing differ in autism.” Authors: Nathaniel J. Zuk, Jaroslav Y. Slobodskoy-Plusnin, Yarden Weiss, VishnuPriya Sampath, Kira M. Düsterwald, Athena Akrami and Merav Ahissar.
- DOI: 10.1523/JNEUROSCI.0315-26.2026
Abstract
Neurocognitive characteristics of perceptual feedback processing differ in autism
Autistic individuals often show reduced adaptation to changing environmental demands, but the neurocognitive mechanisms underlying this difference remain unclear. We asked whether it may be explained by reduced sensitivity to external feedback and changes in task difficulty.
Autistic and non-autistic adults of either sex, matched for cognitive ability, performed a pitch discrimination task in which their response was immediately followed by feedback. Accuracy was similar between the two groups, yet autistic participants exhibited a markedly reduced feedback-related negativity (FRN), an event-related EEG potential differentiating explicit negative versus positive feedback. They also showed weaker modulation of feedback-evoked EEG responses by perceptual difficulty.
Lastly, they showed reduced trial-by-trial behavioral updating of perceptual bias following recent correct and easy trials. These effects were replicated in a follow-up experiment conducted 1–2 years later. When combined they classified autistic and non-autistic participants with about 80% accuracy.
Together, our findings indicate that multiple processes for evaluating perceptual decisions and adjusting subsequent behavior differ in autism and likely contribute to slow updating in changing contexts.
