← Explore

Enhanced Episodic Memory Encoding Following Modulation of the Amygdala via Transcranial Low-intensity Focused Ultrasound

Manoj K. Doss, Sydney R. Lambert, Jason Samaha, Sai Prashanth Raja Sundaram, Charles B. Nemeroff, Gregory A. Fonzo, Joseph E. Dunsmoor

Journal of Cognitive Neuroscience 2026, 1-14 · 10.1162/jocn.a.2687

human healthyhealthybehaviour

Abstract

The amygdala is considered crucial to the formation of emotional episodic memories, yet causal evidence in humans is limited due to challenges in noninvasive neuromodulation of deep brain structures. In a double-blind, sham-controlled, repeated-measures study, we tested whether transcranial low-intensity focused ultrasound (tFUS) targeting the left amygdala before encoding emotional and neutral pictures impacted memory for these pictures 24 hr later (n = 23). Using a putative inhibitory tFUS protocol previously shown to attenuate amygdala BOLD-dependent signal, we hypothesized that preencoding amygdala inhibition would impair emotional memory. Contrary to this hypothesis, active (vs. sham) sonication enhanced-or tended to enhance-memory for both neutral and emotional memory across two memory tests. As a secondary measure of amygdala modulation, participants completed a dynamic emotional facial recognition task (n = 20). Active tFUS selectively enhanced recognition of subtle fearful expressions. Computational modeling further supported tFUS-mediated memory and emotion recognition enhancements. Assuming that this tFUS protocol did indeed inhibit amygdala function, these findings suggest that rather than solely amplifying emotional salience, the amygdala may act as a filter that suppresses weaker, ambiguous signals. Inhibiting the amygdala may thus attenuate this gating function, thereby unmasking subtle affective cues and enhancing their processing into memory. Nevertheless, the purported inhibitory nature of this tFUS protocol requires further investigation. Should these findings replicate, they highlight the potential for tFUS as a translational tool to modulate affective processes for disorders such as depression and posttraumatic disorder characterized by impairments in emotional processing and memory.

Abstract via europepmc.

Specieshuman
Subjects28 participants
Sessions per subject1
Randomisednot reported
Blindingdouble
Sham / controlinactive transducer
Auditory controlnot reported
Readout timingoffline
Anaesthesianot applicable
ReadoutsbehaviourCued recollection and picture recognition memory tests (hit/false alarm rates, DPSD recollection/familiarity modeling); dynamic emotional facial expression (fear/disgust) recognition task
Direction of effectexcitatoryContrary to the hypothesized inhibitory effect of the amygdala protocol, active (vs. sham) sonication enhanced or tended to enhance hit rates, high-confidence memory accuracy/confidence, and recollection/familiarity estimates for neutral and emotional pictures, and enhanced recognition of subtle fearful facial expressions.
Adverse eventsnot reported

Exposures

Exposure 1: tFUS to left amygdala before encoding

Target: basolateral amygdala — “left amygdala (greatest modeled effect in anterior basolateral amygdala)
Device: BrainSonix · Brainsonix · Pulsar 1002

Pulse timing
Waveformpulsed
Fundamental frequency (kHz)650✓✓
Pulse duration (ms)5✓✓
Pulse repetition frequency (Hz)10✓✓
Duty cycle (%)5pulse duration × PRF gives 5%✓✓
Sonication duration (s)30✓✓
Pressure and intensity, by domain
Free-field pressure (kPa)not reported
Free-field Isppa (W/cm²)not reported
Free-field Ispta (W/cm²)not reported
In-situ estimatederatingsingle value
In-situ pressure (kPa)640✓✓
In-situ Isppa (W/cm²)14.4✓✓
In-situ Ispta (W/cm²)0.72✓✓
Protocol, in the paper’s words

Sonication was delivered as 10 min total in 30-sec on/off blocks (continuous pulsing at 10-Hz PRF, 5% duty cycle during each 30-sec 'on' period) immediately before the encoding phase of the emotional episodic memory task.

Flags from extraction

  • randomisedPaper describes a double-blind, sham-controlled, within-subject design with counterbalanced sonication order, but does not explicitly state that arm order was randomized (only 'counterbalanced').
  • auditory_controlPaper states sonication parameters are 'imperceptible' and the sham pad contains a foam layer, but does not describe any explicit auditory masking method to control for audible confounds.
  • exposures[0].timing.sonication_duration_sSonication is delivered in 30-sec on/off blocks over a total 10-min session; the 30-sec 'on' block is treated as the length of one uninterrupted pulse train, with the overall session structure described in protocol_description.