Search Abstracts | Symposia | Slide Sessions | Poster Sessions
Closed-Loop Striatal Stimulation Expands Neural Population Encoding and Enhances Swahili Learning
Poster Session F, Friday, October 2, 2:45 - 4:45 pm, Wangari Maathai
Nathaniel Killian1,2, Emad Eskandar1,2; 1Albert Einstein College of Medicine, 2Montefiore
The use of language is essential for virtually all social interactions. Yet, the neurophysiological mechanisms underpinning language learning and semantic competition are poorly understood, limiting our ability to treat disorders such as aphasia, which affects over 2 million people in the US. Beyond speech therapy, there are no accepted pharmacological or neuromodulatory treatments for chronic aphasia. Past work has shown that high-frequency stimulation of the nucleus accumbens (NAc) during stimulus presentation, coupled with caudate (Cd) stimulation following correct responses, enhances associative learning rates. Because language is acquired primarily through associative learning, we hypothesized that precisely timed stimulation of these striatal hubs would causally enhance language learning. Five epilepsy patients undergoing stereotactic electroencephalography (SEEG) monitoring performed a four-alternative forced-choice Swahili word-learning task with bilateral NAc+Cd stimulation (8 to 16 words per subject were learned over 240-330 trials). Stimulation improved learning rate, approximately doubling on average, within a single ~30-minute session with interleaved stimulated and unstimulated words. We then explored the potential mechanisms underlying the dramatic improvement in learning. First, we found that post-stimulation LTP-like plasticity developed across learning in the NAc during word reading and later in the Cd after correct-response feedback. These plasticity effects accumulated across learning trials and were accompanied by similar changes in the middle temporal gyrus and the inferior frontal gyrus. Using a matched-channel laterality analysis, we found that right-hemisphere ensemble activity represented word identity significantly better with stimulation. Stimulation resulted in a right hemisphere decoding gain of up to 40% early in learning (p<0.001, cluster-based permutation test). Single-trial local field potentials projected into a shared principal component space showed that stimulated words occupied distinct, well-separated trajectories, whereas unstimulated words clustered in a narrower manifold. Pairwise trajectory distances were up to 2.7x larger for stimulated words, and this distinctness was correlated with learning rate (Spearman’s rho = 0.58, p<0.001). Furthermore, the word trajectories followed a consistent anatomical pattern across stimulated and unstimulated words (Kendall’s W, p<0.001, N=5/5 subjects), demonstrating that striatal stimulation created word-specific neural signatures on a stable functional structure, rather than altering connectivity or inducing uniform modulation of activity. Taken together, these results support precise NAc+Cd stimulation for circuit-based language rehabilitation.
Topic Areas: Speech-Language Treatment, Meaning: Lexical Semantics