
# Research Plan

## Problem

We aim to investigate how intracranial theta burst stimulation (TBS) of the basolateral amygdala (BLA) affects neuronal activity at the single-cell level in humans. Previous research has demonstrated that intracranial TBS of the BLA can enhance declarative memory, likely by modulating hippocampal-dependent memory consolidation processes. The amygdala's extensive connectivity to distributed cortical and subcortical regions involved in cognition makes it an ideal target for understanding how electrical stimulation influences neural circuits.

While emerging evidence suggests that intracranial TBS may enhance memory specificity, evoke theta-frequency oscillations, and facilitate short-term plasticity in local field potential recordings, the extent to which exogenous TBS modulates activity at the single-cell level remains poorly understood. This knowledge gap is critical because individual neurons are thought to be the substrate of memory encoding and retrieval.

We hypothesize that intracranial TBS of the BLA will modulate spiking activity within highly connected regions (particularly the hippocampus) and improve memory task performance. Given the amygdala's well-established connectivity to the hippocampus, orbitofrontal cortex, and anterior cingulate cortex, we predict that BLA stimulation will modulate neuronal activity in these sampled regions. We also hypothesize that recorded units will predominantly exhibit enhanced spiking in response to intracranial TBS, based on recent studies reporting enhanced neural plasticity following repetitive direct electrical stimulation.

## Method

We will conduct simultaneous microelectrode recordings from prefrontal and medial temporal structures during a memory task in which intracranial TBS is applied to the BLA. Our approach involves recording from patients with medically refractory epilepsy who are undergoing stereoelectroencephalography for clinical purposes.

To characterize neuronal modulation, we will contrast trial-averaged, peri-stimulation firing rates across stimulation and no-stimulation conditions. We will create peri-stimulation epochs (1 s pre-trial inter-stimulus interval, 1 s after image onset, 1 s during stimulation/after image offset, 1 s post-stimulation) and perform firing rate contrasts across trials within distinct conditions.

Our methodology will employ permutation-based statistical testing, comparing empirical test statistics against null distributions generated by shuffling epoch labels. We will use Wilcoxon signed-rank tests to compare spike counts and control for false positives through permutation testing.

We will classify neurons based on their location, baseline activity, and direction of effect (enhancement vs. suppression). Additionally, we will perform population analyses using linear dimensionality reduction with PCA to examine temporal dynamics of dominant modes of neuron coactivity in low-dimensional subspace.

## Experiment Design

We will record single-unit activity from patients with medically refractory epilepsy as they complete a visual recognition memory task. The experimental design consists of an encoding session where neutral valence images are presented, followed by a self-paced retrieval session approximately 24 hours post-encoding.

During the encoding session, patients will receive either 80 or 160 trials of bipolar intracranial TBS to contiguous macroelectrode contacts in the BLA. An equal number of "no-stimulation" trials will be randomly interspersed to evaluate stimulation effects on memory performance and control for neuronal modulation resulting from experimental stimuli.

We will deliver charge-balanced, bipolar, biphasic rectangular pulses over a 1 s period with a 50% duty cycle. Stimulation pulses will be delivered at 50 Hz and nested within eight equally-spaced bursts (~8 Hz). We will use 1 mA current amplitude as the primary parameter, with a subset of experiments using 0.5 mA with variable pulse frequencies.

Neurophysiological data will be recorded using neural signal processors sampling at 30 kHz. We will use Behnke-Fried depth electrodes containing both macro- and microelectrode contacts for recording local field potentials and extracellular action potentials.

For spike detection and sorting, we will filter microelectrode data between 250-500 Hz and re-threshold offline at -3.5 times the root mean square of the signal. Units will be isolated using semi-automated processes applying T-distribution expectation maximization method on principal components of detected waveforms.

We will exclude units with trial-averaged baseline firing rates < 0.1 Hz from analyses due to limited ability to detect modulation robustly. Memory performance will be assessed using d-prime calculations, and we will test associations between neuronal modulation and behavioral outcomes using linear mixed-effects models.