← Explore

Interaction and functional specialization across a distributed neural circuit for flexible task control in macaques

K. Marche, N. Trudel, U. Schüffelgen, S. Smart, C. M. Harbison, J. Scholl, J. Sallet, F-X Neubert, J. Algermissen, N. Khalighinejad, JX O’Reilly, MC Klein-Flügge, MFS Rushworth

Nature Communications 2026 · 10.1038/s41467-025-65766-0

nonhuman primatehealthybehaviour

Abstract

Reversal tasks have been regarded as probes of behavioural inhibition and linked to prefrontal and specifically orbitofrontal cortex. The centrality of behavioural inhibition to reversal task performance and the task's dependence on particular prefrontal sub-regions have, however, been questioned in primates. Using a combination of whole brain recording, transient ultrasonic disruption, two types of reversal task, and a task model emphasizing identification of transitions between latent states, we show that male macaques track latent state transitions in addition to choice values in reversal tasks. Activity reflecting both these features is prominent in dorsomedial frontal cortex, and anterior and dorsomedial thalamus when, and just before, animals select choices. By contrast, hippocampal activity continually tracks the probability of a reversal between latent states. We identify patterns of activity interaction spanning the three nodes of this circuit and demonstrate that disruption of each leads to reversal task impairment albeit in different ways.

Abstract via europepmc.

Speciesrhesus macaque (Macaca mulatta)
Subjects3 animals
Sessions per subject5, 6swept
Randomisedyes
Blindingnot reported
Sham / controlinactive transducer, sound only
Auditory controlsound only sham
Readout timingoffline
Anaesthesiaawake
Readoutsbehaviour
Direction of effectmixed or unclearTUS to dmPFC, hippocampus, and anterior/dorsomedial thalamus each significantly reduced the influence of the model-derived reversal estimate on choice switching relative to sham, but in different ways: dmPFC TUS abolished the influence of nearly all decision variables (reversal estimate, uncertainty) leaving only recent reward history; hippocampus TUS selectively removed the influence of the reversal estimate; thalamus TUS reversed the sign of the reversal estimate's influence on switching.
Adverse eventsnot reported

Exposures

Exposure 1: dmPFC TUS

Target: dorsomedial prefrontal cortex — “dmPFC
Device: NeuroFUS · Sonic Concepts (sold by Brainbox) · NeuroFUS CTX-250

Pulse timing
Waveformpulsed
Fundamental frequency (kHz)250✓✓
Pulse duration (ms)30✓✓
Pulse repetition frequency (Hz)10✓✓
Duty cycle (%)30pulse duration × PRF gives 30%✓✓
Sonication duration (s)40✓✓
Pressure and intensity, by domain
Free-field pressure (kPa)not reported
Free-field Isppa (W/cm²)60✓✓
Free-field Ispta (W/cm²)not reported
In-situ estimatesimulationsingle value
In-situ pressure (kPa)not reported
In-situ Isppa (W/cm²)16✓✓
In-situ Ispta (W/cm²)not reported
Protocol, in the paper’s words

TUS was delivered bilaterally (40 s to the left hemisphere, 40 s to the right hemisphere) using a duty cycle of 30%, pulse length of 30 ms, and pulse repetition interval of 100 ms. This was an offline protocol: monkeys began the behavioural task 5-10 minutes after stimulation ended. Each stimulation condition (dmPFC, hippocampus, thalamus, or sham) was repeated 5-8 times per animal on separate days, at least 24 hours apart, with a sham day scheduled at the start and end of the series.

Exposure 2: Hippocampus TUS

Target: hippocampus — “hippocampus
Device: NeuroFUS · Sonic Concepts (sold by Brainbox) · NeuroFUS CTX-250

Pulse timing
Waveformpulsed
Fundamental frequency (kHz)250✓✓
Pulse duration (ms)30✓✓
Pulse repetition frequency (Hz)10✓✓
Duty cycle (%)30pulse duration × PRF gives 30%✓✓
Sonication duration (s)40✓✓
Pressure and intensity, by domain
Free-field pressure (kPa)not reported
Free-field Isppa (W/cm²)60✓✓
Free-field Ispta (W/cm²)not reported
In-situ estimatesimulationsingle value
In-situ pressure (kPa)not reported
In-situ Isppa (W/cm²)18.5✓✓
In-situ Ispta (W/cm²)not reported
Protocol, in the paper’s words

TUS was delivered bilaterally (40 s to the left hemisphere, 40 s to the right hemisphere) using a duty cycle of 30%, pulse length of 30 ms, and pulse repetition interval of 100 ms. This was an offline protocol: monkeys began the behavioural task 5-10 minutes after stimulation ended. Each stimulation condition (dmPFC, hippocampus, thalamus, or sham) was repeated 5-8 times per animal on separate days, at least 24 hours apart, with a sham day scheduled at the start and end of the series.

Exposure 3: Thalamus TUS

Target: anterior thalamic nucleus, mediodorsal nucleus — “anterior and dorsomedial thalamus
Device: NeuroFUS · Sonic Concepts (sold by Brainbox) · NeuroFUS CTX-250

Pulse timing
Waveformpulsed
Fundamental frequency (kHz)250✓✓
Pulse duration (ms)30✓✓
Pulse repetition frequency (Hz)10✓✓
Duty cycle (%)30pulse duration × PRF gives 30%✓✓
Sonication duration (s)40✓✓
Pressure and intensity, by domain
Free-field pressure (kPa)not reported
Free-field Isppa (W/cm²)60✓✓
Free-field Ispta (W/cm²)not reported
In-situ estimatesimulationsingle value
In-situ pressure (kPa)not reported
In-situ Isppa (W/cm²)17✓✓
In-situ Ispta (W/cm²)not reported
Protocol, in the paper’s words

TUS was delivered bilaterally (40 s to the left hemisphere, 40 s to the right hemisphere) using a duty cycle of 30%, pulse length of 30 ms, and pulse repetition interval of 100 ms. This was an offline protocol: monkeys began the behavioural task 5-10 minutes after stimulation ended. Each stimulation condition (dmPFC, hippocampus, thalamus, or sham) was repeated 5-8 times per animal on separate days, at least 24 hours apart, with a sham day scheduled at the start and end of the series.

Flags from extraction

  • n_sessions_per_subjectPaper states 'each stimulation condition was repeated 5-8 times for each animal' (per target/sham condition, not a single total-sessions count); recorded as the stated range.
  • randomisedPaper describes the order of TUS targets as 'pseudo-randomized' rather than fully randomised; recorded as true.
  • auditory_controlSham TUS matched the real procedure except that 'sonication was not triggered although the TUS sound was simulated'; classified as sound_only_sham.
  • exposures[2].target.termsPaper refers to the thalamic target variably as 'anterior medial thalamus' and 'anterior and dorsomedial thalamus'; both anterior_thalamic_nucleus and mediodorsal_nucleus terms are given.
  • direction_of_effectTUS effects differed qualitatively across the three targets (loss of influence vs. reversal of influence direction), so classified as mixed_or_unclear.