Head movement allocation in freely-moving macaques learning object-context sequences

Two macaques learned numerous unique conditional object-context sequences across four spatiotemporal contexts and recalled correct choices after weeks without exposure, while head kinematics were decoded during choice.

• Why it matters: Understanding how primate brains link to changing environments requires tasks that constrain behavior minimally yet provide conditional cues that drive varied dynamics.
• What they did: Freely moving monkeys used a multi-level Treehouse with independently controlled touchscreens; head kinematics were tracked and recurrent neural networks decoded object position, sequence features, and occupied screen stations during the choice epoch.
• The result: One model predicted the object about to be touched, including on error trials, and tracked gaze; other models decoded sequence corner, ordinal step, and station, while uncertainty biased initial search toward closest targets and training improved movement smoothness without accuracy change.

AI summary of the abstract · may contain errors · not medical advice · read the paper.