A causal role of the NMDA receptor in recurrent processing during perceptual integration.

Summary

Perceptual inference—how the brain integrates visual features into a coherent whole—depends on recurrent processing, the back-and-forth communication between higher and lower cortical regions. Animal studies have implicated NMDA receptors in this process, but human evidence was lacking. In two double-blind, placebo-controlled experiments with healthy participants, the NMDA receptor antagonist memantine selectively improved the brain's ability to decode complex visual illusions (Kanizsa triangles) that require recurrent processing, while leaving simpler visual processing (contrast and collinearity) unaffected. This enhancement occurred only when stimuli were attended and consciously perceived. The findings suggest that blocking NMDA receptors can enhance recurrent processing for attended objects, linking animal and human research on the neural basis of conscious perception.

Study at a glance

Abstract

Perceptual inference requires the integration of visual features through recurrent processing, the dynamic exchange of information between higher- and lower-level cortical regions. While animal research has demonstrated a crucial role of NMDA receptors in recurrent processing, establishing a causal link between NMDA receptors and recurrent processing in humans has remained challenging. Here, we report two pharmacological studies with randomized, double-blind, crossover designs in which we administered the NMDA antagonist memantine, while collecting human electroencephalography (EEG). We trained and tested EEG classifiers to reflect the processing of specific stimulus features with increasing levels of complexity, namely differences in stimulus contrast, collinearity between local line elements, and illusory surfaces of a Kanizsa triangle. In two experiments involving different participants and visual tasks, we found that memantine selectively improved decoding of the Kanizsa illusion, known to depend on recurrent processing, while leaving decoding of contrast and collinearity largely unaffected. Interestingly, the results from an attentional blink (experiment 1) and task-relevance manipulation (experiment 2) showed that memantine was only effective when the stimulus was attended and consciously accessed. These findings suggest that NMDA inhibition through memantine enhances recurrent processing, especially for attended objects, and thereby provide a crucial step toward bridging animal and human research, shedding light on the neural mechanisms underpinning perceptual inference and conscious perception.