State-dependent associative plasticity highlights function-specific premotor-motor pathways crucial for arbitrary visuomotor mapping

Abstract

Arbitrary visuomotor mapping (AVMM) showcases the brain's ability to link sensory inputs with actions. The ventral premotor cortex (PMv) is proposed as central to sensorimotor transformations, relaying descending motor commands through the primary motor cortex (M1). However, direct evidence of this pathway's involvement in AVMM remains elusive. In four experiments, we used cortico-cortical paired associative stimulation (ccPAS) to enhance (ccPAS_PMv-M1) or inhibit (ccPAS_M1-PMv) PMv-to-M1 connectivity via Hebbian plasticity. Leveraging state-dependent properties of transcranial magnetic stimulation, we targeted function-specific visuomotor neurons within the pathway, testing their physiological/behavioral relevance to AVMM. State-dependent ccPAS_PMv-M1, applied during motor responses to target visual cues, enhanced neurophysiological and behavioral indices of AVMM, while ccPAS_M1-PMv had an opposite influence, with the effects being more pronounced for target relative to control visual cues. These results highlight the plasticity and causal role of spatially overlapping but functionally specific neural populations within the PMv-M1 pathway in AVMM and suggest state-dependent ccPAS as a tool for targeted modulation of visuomotor pathways.

Fig. 1.. Experiment 1 methodology.

(A) General design. We assessed CSE, SICI, and ICF in three separate test blocks before and after state-dependent ccPAS. In two different sessions, participants were submitted to ccPAS_PMv-M1 and ccPAS_M1-PMv during an AVMM task. (B) Plasticity induction procedure. State-dependent ccPAS paired pulses were administered 500 ms after the onset of the target visual cue while participants were responding with a finger movement involving the target muscle (e.g., the FDI muscle in response to the blue square). During the control visual cue, requiring a movement involving the control muscle no TMS pulses were delivered. The two cues appeared alternatively, for a total of 90 trials for each color. (C) Individual targeted cortical sites reconstructed on a standard template using MRIcron software (MRIcron/NPM/dcm2nii) after conversion to MNI space for illustrative purposes. (D) Test block procedure. TMS pulses were administered over M1 to induce MEPs at 250 to 320 ms after the onset of the target and control visual cues and assess CSE, SICI, and ICF.

Fig. 2.. Results of experiment 1, testing the influence of state-dependent ccPAS.

(A) Following ccPAS_PMv-M1, CSE increased in the target muscle, and the facilitatory effect was more prominent for the target visual cue (red line). (B) No CSE change was observed in the control muscle. (C) After ccPAS_M1-PMv, CSE decreased in the target muscle, with a more pronounced inhibitory response to target visual cues (blue line). (D) No CSE change was observed in the control muscle. Lighter circles and lines denote individual data points; darker squares and lines denote group means. Box plots represent the median, first and third quartiles, and the 95% confidence interval of individual data points. Asterisks indicate significant post hoc comparison: *P < 0.05 and ***P < 0.001.

Fig. 3.. Methodology and results of experiments 2 and 3.

(A to C) Experiment 2 testing the influence of AVMM without ccPAS. (A) Design. (B) Individual targeted sites reconstructed on a standard template using MRIcron software (MRIcron/NPM/dcm2nii) after conversion to MNI space for illustrative purposes. (C) Results: Following the AVMM task, CSE increased in both muscles and irrespective of the presented visual cue. (D to F) Experiment 3 testing the influence of state-dependent ccPAS_Sham-M1. (D) Design. (E) Individual targeted cortical sites. (F) Results: State-dependent ccPAS_Sham-M1 increased CSE in the target muscle only, irrespective of the presented visual stimulus. Lighter circles and lines denote individual data points; darker squares and lines denote group means. Box plots represent the median, first and third quartiles, and the 95% confidence interval of individual data points. *P < 0.05, and ***P < 0.001.

Fig. 4.. Summary of results of experiments 1 to 3.

LTP-like enhancement of visuomotor specificity following state-dependent ccPAS_PMv-M1 (red), LTD-like visuomotor-specific effect following state-dependent ccPAS_M1-PMv (blue), nonspecific CSE increase following the AVMM task with no TMS (green), and motor-specific LTP-like effect following state-dependent M1 stimulation (yellow). Histograms represent the mean of each condition; error bars represent 1 SEM. **P < 0.01 and ***P < 0.001.

Fig. 5.. Results of experiment 1, testing the influence of state-dependent ccPAS on intracortical excitability indices.

(A) No changes were observed for SICI following ccPAS_PMv-M1. (B) ccPAS_PMv-M1 increased ICF in response to the target visual cue. (C) ccPAS_M1-PMv increases the magnitude of SICI irrespective of the presented visual cue. (D) No changes were observed for the ICF index. Lighter circles and lines denote individual data points; darker squares and lines denote group means. Box plots represent the median, first and third quartiles, and the 95% confidence interval of individual data points. spTMS, single pulse TMS. Asterisks indicate significant post hoc comparison: *P < 0.05.

Fig. 6.. Methodology and results of experiment 4 testing the influence of state-dependent ccPAS on AVMM performance.

(A) Design. (B) AVMM training—task. Participants were instructed to respond to two visual cues (one target and one control) with their index finger and to two different visual cues (one target and one control) with their thumb finger. The four cues appeared randomly, for a total of 90 trials for each color (see Materials and Methods for details). During the state-dependent ccPAS protocol, paired pulses were administered at the onset of target movements (i.e., during motor responses to target cues). No TMS pulses were delivered during control movements (i.e., during responses to control cues). (C) Individual targeted sites reconstructed on a standard template using MRIcron software (MRIcron/NPM/dcm2nii) after conversion to MNI space for illustrative purposes. (D) d′ performance in the ccPAS_Sham-M1 group did not change over time when responding to either the target or the control color. (E) ccPAS_PMv-M1 and ccPAS_M1-PMv induced opposite effects on d′ selectively in responding to target stimuli but not control ones. (F) Individual data points of d′ modulation in responding to target stimuli in the ccPAS_PMv-M1 and ccPAS_M1-PMv sessions. Error bars represent one SD; box plots represent the median, first and third quartiles, and the 95% confidence interval of individual data points. *P < 0.05 and **P < 0.01.