Shervin Rahimpour

The paper investigates network competition in the brain using electrophysiology and neuroimaging, finding temporally layered antagonism between default and task-related networks, with distinct patterns in the hippocampus.

The paper uses deep reinforcement learning to model risk-adaptive decision-making strategies and identifies low-dimensional neural dynamics associated with high-performing strategies in humans.

This paper identifies distinct neural mechanisms for speech planning and execution using intracranial recordings.

This paper shows that speech BCIs can be trained on combined patient data by aligning neural data to a shared latent space, improving decoding performance compared to patient-specific models.

The paper investigates neural representations of time during decision-making under risk using human intracranial electro-physiology and a temporal convolution network to predict response timing.

This study uses direct electrical stimulation of the human amygdala to show it enhances object memory but not scene memory, suggesting specificity in memory prioritization.

This study shows that human hippocampal neurons use polysemantic coding, where neurons represent multiple unrelated semantic features through high-dimensional superposition, enabling efficient contextualization of meaning.

The paper shows that semantic features in word meaning are encoded in the brain through geometric structures similar to those in language models.

This study shows that an asymmetric reinforcement learning model better explains human decision-making under risk than symmetric models.

This study shows that aperiodic exponents from intracranial EEG can differentiate depression symptom severity and correlate with specific brain regions and networks.

The paper investigates neural bursting activity in human prefrontal and temporal brain areas during visual working memory and finds that gamma and beta bursting correlates with memory performance and is coupled via a phase-burst code.

This study examines how theta burst stimulation of the basolateral amygdala modulates single-neuron activity in humans during a memory task, finding that ~30% of neurons show heterogeneous responses.

This study examines how human neurons respond to single-pulse electrical stimulation, finding that most neurons show firing rate suppression, with differences based on cell type and stimulation distance.

This study uses human brain recordings to show that creative thinking is associated with distinct neural states in the default mode and dorsal attention networks.

This study shows that spreading depolarizations (SDs) are associated with seizure termination in both human and rodent models, with SDs reducing seizure duration and delaying subsequent events.

The paper investigates how neuronal representations of reward and risk expectations differ between impulsive and less impulsive decision-makers during a risky choice task.

This study improves speech decoding accuracy using high-resolution ECoG recordings, achieving better phoneme decoding than standard methods.

This study uses intracranial recordings to analyze how neural activity correlates with binary and graded pain responses during tonic thermal stimuli in humans.