Task relevance amplifies neural encoding of temporal prediction errors during fast sensorimotor behavior

MEG DATA FOR THE ROLE OF TASK RELEVANCE IN THE NEURAL ENCODING OF TEMPORAL PREDICTION ERRORS Associated manuscriptTask relevance amplifies neural encoding of temporal prediction errors during fast sensorimotor behaviorDavide Sometti, Ileana Rossetti, Cornelius Schwarz, Markus Siegel, Christoph Braun and Antonino GrecoCommunications Biology CONTENTS AND SCOPE This dataset contains preprocessed, source-reconstructed MEG data and behavioral information from 11 participants performing a tactile sensorimotor task, investigating how task relevance shapes the neural encoding of temporal prediction errors during fast sensorimotor behavior. Using Magnetoencephalography (MEG), computational modeling and multivariate decoding, we examined how deviations from expected tactile stimulus timing are represented when participants respond to stimulation versus passively perceive it, and whether these representations follow a sensory or motor reference frame. TASK AND CONDITIONS Tactile stimulation consisted of brief pneumatic pulses to an index fingertip. In stimulus-response conditions, participants responded as quickly as possible using a thumb button press. Response-only blocks involved self-paced button presses approximately every 2 to 3 seconds. Lx means left hand and Rx means right hand. In SR names, the first hand is the stimulated hand and the second is the responding hand. SO_Lx: stimulus only, left index finger; stimulus aligned.SO_Rx: stimulus only, right index finger; stimulus aligned.RO_Lx: response only, left thumb; response aligned.RO_Rx: response only, right thumb; response aligned.SR_LxLx: left stimulation, left response; ipsilateral.SR_LxRx: left stimulation, right response; contralateral.SR_RxRx: right stimulation, right response; ipsilateral.SR_RxLx: right stimulation, left response; contralateral. Each SR condition has two fields: suffix _Saligned means time zero is stimulus onset; suffix _Raligned means time zero is response onset. These are two representations of the same retained trials, not additional experimental conditions. SO and RO fields have no alignment suffix because their alignment follows the task. DATA ACQUISITION, PREPROCESSING & SOURCE RECONSTRUCTION MEG was recorded with a 275-sensor CTF system at approximately 585.94 Hz. Continuous signals were high-pass filtered at 1 Hz and low-pass filtered at 30 Hz using sixth-order Butterworth filters. No baseline correction was applied. Trials and channels were visually inspected for artifacts and rejected as appropriate. Independent component analysis was used to identify and remove non-cortical components, including ocular, cardiac and muscle artifacts. Data were resampled to 100 Hz and projected into source space. Single-trial source activity was reconstructed using linearly constrained minimum variance (LCMV) beamforming. Source activity was parcellated into 72 regions using the Desikan-Killiany atlas. DATA STRUCTURE subject.<condition> is a FieldTrip structure with individual trials. Its principal fields are: trial: numeric array, retained trials x brain regions x time samples.time: common time vector in seconds relative to the alignment event.label: region labels, in the same order as dimension 2 of trial.dimord: FieldTrip dimension description, normally rpt_chan_time.trial_id: original trial identifiers, in neural-array order.bhv: bhv contains all original trials, including excluded trials. - trial: original trial number; matched to neural data through trial_id. - stimulus_time, response_time: event timestamps in seconds. - isi: interstimulus interval in seconds; NaN for the first trial. - rt: reaction time in seconds; NaN for SO and excluded trials. - badtrial: true for excluded trials, false for included trials.