henhaobio
Sign in

Neuroscience & Biobehavioral ReviewsNeurosciencePeer-reviewed journal article

Neural Correlates of Motor Imagery, Action Observation, and Movement Execution: A Comparison Across Quantitative Meta-Analyses

Robert M. HardwickiD, Svenja CaspersiD, Simon Bodo EickhoffiD, Stephan Patrick SwinneniD

PDFDOIPreprint

In-depth summary

Claude Opus 5.5 · xhigh effort · 8 Nov 2026 · from the full text of version 1 (10 Oct 2017) · CC BY 4.0

Coordinate-based meta-analyses reveal a premotor, rostral parietal, and somatosensory core shared by motor imagery, action observation, and execution; only imagery and execution share putamen and cerebellum.

Earlier reviews judged the overlap between motor imagery, action observation, and movement execution subjectively, and prior meta-analyses mapped the imagery and observation networks separately without comparing them quantitatively to each other or to execution. Here, activation likelihood estimation (ALE) contrasts, volume overlaps, and conjunctions test which regions are consistently shared, which matters for neurofeedback and rehabilitation uses of action simulation.

Key results

  1. ShownOnly a premotor network (pre-SMA, SMA-proper, dorsal and ventral premotor cortex), a rostral inferior parietal region (area PFt), and somatosensory regions converged across all three tasks. Primary motor cortex was consistently recruited only by movement execution.

    A grand minimum-statistic conjunction (clusters ≥100mm^3) found bilateral PMd, PMv, pre-SMA, left SMA-proper, bilateral PFt, and sensorimotor clusters. Voxels shared by all three tasks made up 14% of the imagery volume, 10% of the observation volume, and 19% of the execution volume. · 3.7 Motor Imagery ∩ Action Observation ∩…; 4.2.2 Parietal Cortex; 3.4.4 Imagery, Observation, and Execution; 4.1.3 Movement Execution

  2. ShownSubcortical recruitment separated the tasks. Imagery and execution both consistently engaged the putamen and cerebellar lobule VI, while action observation produced no subcortical peak maxima. Apparent cerebellar clusters in conjunctions with observation came from adjacent extrastriate clusters.

    The imagery ∩ execution conjunction included bilateral putamen and cerebellar lobule VI, and putamen and cerebellum favored imagery over observation. Observation clusters dipped into the cerebellum only slightly, with no peak maxima there; the cerebellar cluster in the imagery ∩ observation conjunction traced to an extrastriate observation cluster. · 3.1.2 Action Observation; 3.3.1 Motor Imagery vs Action Observation; 3.5.1 Motor Imagery ∩ Action Observation; 3.5.2 Motor Imagery ∩ Movement execution

  3. ShownVoxel overlap was largest between imagery and observation, intermediate between imagery and execution, and smallest between observation and execution (the putative mirror comparison). The pattern held after volume matching.

    53% of imagery voxels fell inside the observation network, and imagery covered 39% of that network (43% after matching). Imagery vs execution: 29% and 40% (33% after matching). Observation vs execution: 13% and 24% (14% after matching). · 3.4.1 Imagery and Observation; 3.4.2 Imagery and Execution; 3.4.3 Observation and Execution

  4. ShownCompared with observation, imagery more consistently engaged SMA, premotor, parietal, putamen, and cerebellar regions; observation was favored only in small inferior frontal and right parietal clusters. Execution exceeded both simulation tasks in SMA-proper, left M1, and ventral premotor cortex.

    Method: random-effects ALE subtraction with 10,000 label shuffles, posterior probability P>0.95, and clusters ≥100mm^3. The clusters favoring observation (inferior frontal, right inferior/superior parietal) lie where mirror neurons are reported in non-human primates. · 2.5 Analysis Procedure; 3.3.1 Motor Imagery vs Action Observation; 3.3.2 Motor Imagery vs Movement Execution; 3.3.3 Action Observation vs Movement Execution

  5. ShownEffector subanalyses showed clear somatotopy for execution in M1 and the right cerebellum, limited premotor and parietal somatotopy for observation, and little somatotopy for imagery. Observing faces also engaged the left thalamus and bilateral amygdala.

    Execution M1 clusters were ordered superior (leg), central (arm/hand), and inferior (face); right-cerebellum clusters ran anterior, central, and posterior. The exploratory face-imagery set of six studies showed no convergence; the 12 face-execution studies resembled the main network. · 3.2.1 Motor Imagery by Body Part; 3.2.2 Action Observation by Body Part; 3.2.3 Movement Execution by Body Part

Methods, data and reproducibility

Revised activation likelihood estimation (ALE) meta-analyses of whole-brain fMRI/PET coordinates from healthy adults (cluster-level FWE p<0.05). These were followed by leg, arm, and face subanalyses, permutation-based ALE subtraction contrasts, volume-matched voxel overlap comparisons, and minimum-statistic conjunctions across imagery, observation, and execution.

  • Motor imagery fMRI/PET experiments (whole-brain MNI/Talairach coordinates, within-subject contrasts) from a PubMed search run up to June 2017 · human (healthy adults ≥18 years) · n = 487 search hits; 205 papers inspected in full
  • Action observation fMRI/PET experiments, including observation before imitation (search terms covered mirror neurons and imitation) · human (healthy adults ≥18 years) · n = 784 search hits; 417 papers inspected in full
  • Movement execution experiments taken from control conditions in the imagery and observation papers · human (healthy adults ≥18 years)
  • Exploratory face motor imagery subset · human · n = six studies
  • Face movement execution subset · human · n = 12 studies
Methods & tools
Activation likelihood estimation (ALE), revised algorithm, Random-effects ALE subtraction contrast analysis, Minimum-statistic conjunction analysis, Lancaster transform (Talairach to MNI), SPM Anatomy Toolbox 2, Human Motor Area Template (HMAT; Mayka et al. 2006), PubMed, fMRI, PET
Reagents & environment
Revised ALE algorithm (no software version or implementation named); anatomical labeling with the SPM Anatomy Toolbox 2 extension and the Mayka et al. (2006) HMAT template.
Reproducibility gaps
No code or data repository links are given. The list of included experiments and the peak coordinates are in Table 1, Tables 2 to 8, and Supplementary Table 1, not in a public archive.
Limitations (6)
  • The execution meta-analysis drew on relatively few studies. Its tasks were also relatively simple, which the authors suggest may explain why parietal involvement was weaker in execution than in imagery or observation. · stated by the authors
  • BOLD and PET hemodynamic signals give only indirect evidence of functional equivalence, because shared voxels may reflect separate neuronal subpopulations. Cross-modal repetition-suppression studies that searched only premotor regions could not be included. · stated by the authors
  • During action simulation, M1 may be active below the level needed to produce peak maxima. The lack of consistent M1 convergence therefore does not rule out weaker M1 engagement. · stated by the authors
  • Cerebellar clusters in conjunctions involving action observation may come from extrastriate visual clusters that happen to sit next to the cerebellum and extend slightly into it, rather than from shared cerebellar recruitment. · stated by the authors
  • Experiment counts differed across tasks, and execution data came only from control conditions. No publication-bias or single-experiment influence analysis is reported, so the robustness of the smaller networks and the six-study face-imagery subset is uncertain. · noted in this summary
  • Observation before imitation was pooled with passive observation, and results were not split by intention to imitate, even though work the authors cite links M1 recruitment during observation to imitation intent. · noted in this summary

Read this if you use motor imagery or action observation for neurofeedback, brain-computer interfaces, or stroke rehabilitation and need quantitative evidence of which motor regions each task shares with movement execution.

Citations
906
FWCI
93.97
Field percentile
top 1%

Abstract

There is longstanding interest in the relationship between motor imagery, action observation, and movement execution. Several models propose that these tasks recruit the same brain regions in a similar manner; however, there is no quantitative synthesis of the literature that compares their respective networks. Here we summarized data from neuroimaging experiments examining Motor Imagery (303 experiments, 4,902 participants), Action Observation (595 experiments, 11,032 participants), and related control tasks involving Movement Execution (142 experiments, 2,302 participants). Motor Imagery recruited a network of premotor-parietal cortical regions, alongside the thalamus, putamen, and cerebellum. Action Observation involved a cortical premotor-parietal and occipital network, with no consistent subcortical contributions. Movement Execution engaged sensorimotor-premotor areas, and the thalamus, putamen, and cerebellum. Comparisons across these networks highlighted key differences in their recruitment of motor cortex,and parietal cortex, and subcortical structures. Conjunction across all three tasks identified a consistent premotor-parietal and somatosensory network. These data amend previous models of the relationships between motor imagery, action observation, and movement execution, and quantify the relationships between their respective networks.\n\nHighlightsO_LIWe compared quantitative meta-analyses of movement imagery, observation, and execution\nC_LIO_LISubcortical structures were most commonly associated with imagery and execution\nC_LIO_LIConjunctions identified a consistent premotor-parietal-somatosensory network\nC_LIO_LIThese data can inform basic and translational work using imagery and observation\nC_LI

Reproducibility ledger

Nobody has reported trying to reproduce this paper yet. Did you replicate an experiment or run the code? Your report saves the next lab a week.

Sign in to file a reproducibility report.

Related papers

Similarity Map
SelectedSame fieldCross-field
THISCaenegem '24Martel '25Dekleva '24Ogawa '21DuPont '23Imai '25
80% semantic matchNeuroscience
Elise E Van Caenegem, Marcos Moreno Verdú, Baptiste M Waltzing+4 more·Neuroscience & Biobehavioral Reviews·Sep 19, 2024· 17 citations