Characterizing the Action-Observation Network Through Functional Near-Infrared Spectroscopy: A Review

The paper addresses how best to study the human action-observation network (AON)—the overlapping systems supporting action execution and observation—given ongoing debates about the human mirror neuron system and the limitations of existing neuroimaging methods. EEG (mu rhythm desynchronization) and fMRI have been widely used but face issues of spatial specificity, motion sensitivity, ecological validity, and often omit critical execution conditions. The review proposes functional near-infrared spectroscopy (fNIRS) as a method well-suited to AON paradigms due to its portability, motion tolerance, and improved spatial resolution over EEG, and evaluates adult AON research using fNIRS to inform methodological standards and future directions, including developmental studies and multimodal approaches.

• EEG and the AON: Mu (8–12 Hz) desynchronization at central sites occurs during both action observation and execution, typically stronger for execution. However, many EEG studies exclude either execution or observation conditions, and mu measures are confounded by alpha and beta rhythms and poor spatial specificity, raising questions about whether effects reflect AON or attention-related processes. High-density/source localization can help but is challenging, especially developmentally.
• fMRI and the AON: Offers higher spatial resolution and has implicated ventral PMC, IPL, STS/SPL, and MFG in AON-related tasks. Yet motion sensitivity limits inclusion of execution conditions (only ~30% included execution in a meta-analysis), reducing the ability to compare execution vs observation, and tasks often lack ecological validity. These constraints limit strong inferences about overlapping AON regions.
• Rationale for fNIRS: Measures HbO/HHb changes with higher tolerance to movement, portability, and suitability across development. It provides better spatial resolution than EEG (though inferior to fMRI) and enables naturalistic, live-action paradigms including motor execution and observation. The review synthesizes findings from 14 adult fNIRS studies that included both execution and observation conditions to determine consistent methodological practices and cortical activations associated with the AON.

This is a critical review of adult fNIRS studies of the AON.

• Search date: July 9, 2019.
• Databases: PubMed, Web of Science, Scopus.
• Search terms: [(near-infrared spectroscopy OR fNIRS OR NIRS OR near-infrared spectroscopy) AND (mirror neuron OR action observation OR action-observation OR action/observation OR execution observation OR execution-observation OR execution/observation)].
• Results: 137 records (64 unique after duplicates).
• Inclusion criteria: (1) peer-reviewed journal; (2) empirical article; (3) used fNIRS to measure brain hemodynamics; (4) contained and analyzed both execute and observe conditions.
• Exclusions: 12 not peer-reviewed; 9 not empirical; 1 without fNIRS data; 28 lacking both execute and observe conditions.
• Final sample: 14 adult fNIRS studies.
• Extracted methodological details included systems used, channel counts, probe placement (uni/bilateral), paradigm design (stimulus type, trial structure), and analytic approaches.
• Summary of systems/coverage: NIRx (5/14), Hitachi (4/14), Shimadzu (4/14), custom (1/14); channels ranged 8–54 (median 24); unilateral probes 9/14 (64.3%), bilateral 5/14 (~35.7%); ROIs commonly included SM1, PMC, IFG, IPL; probe localization and fiducial reporting varied widely.
• Paradigm characteristics: Majority used pre-recorded video stimuli for observation (10/14, 71.4%); event-related designs were common; average 20.21 trials per condition (median 8; range 1–54); intertrial intervals adapted to hemodynamic response timing. Live-action observation was rare despite evidence that live stimuli elicit stronger AON-related responses.

• Feasibility: fNIRS reliably detects cortical activation associated with both action execution and observation, supporting its utility for AON research.
• Hemispheric patterns: Observation often elicited bilateral or ipsilateral activation (e.g., Bhat et al., 2017; Crivelli et al., 2018; Holper et al., 2010). Execution produced stronger activation overall, typically localized to contralateral sensorimotor regions (SM1/PMC) and bilateral premotor areas (e.g., Shimada and Abe, 2010; Balconi and Cortesi, 2016; Balconi et al., 2017; Holper et al., 2010).
• Core regions: Parietal cortex (including IPL/SPL) was frequently activated in both observation and execution relative to baseline, suggesting a central role in the AON across fNIRS studies (Egetemeir et al., 2011; Koehler et al., 2012; Balconi and Cortesi, 2016; Balconi et al., 2017).
• Stimulus modality: Observation with live-action elicits stronger differentiation between action and object observation than video stimuli; reliance on videos (10/14 studies) may attenuate AON-related responses.
• Probe configuration: Unilateral probes predominated (9/14), likely to prioritize coverage of diverse AON-related ROIs. Bilateral probes revealed lateralization differences and broader observation-related activation, indicating the value of bilateral coverage for characterizing network-wide effects.
• Trial design: Event-related designs allowed modeling full hemodynamic responses (e.g., amplitude, time-to-peak). Average trials per condition were modest (mean ~20, median 8), reflecting the need for longer intertrial intervals and SNR considerations.
• Analytic gaps: Many studies assessed condition-wise activation vs baseline without directly comparing execution vs observation magnitudes within regions, limiting precision in identifying overlapping AON nodes.
• Overall: fNIRS findings converge on stronger, more focal execution-related activation in contralateral motor areas and consistent parietal involvement across both conditions, with observation showing broader bilateral effects.

The review’s findings address the central question of whether fNIRS can effectively characterize the AON by demonstrating robust, condition-specific cortical hemodynamics: focal contralateral motor activation during execution, broader bilateral patterns during observation, and consistent parietal involvement across both. fNIRS’s tolerance to motion and portability enable inclusion of ecologically valid execution and live observation conditions that are difficult in fMRI and spatially ambiguous in EEG. Methodological recommendations include bilateral probe coverage to capture lateralization and network-wide effects; use of live-action stimuli to maximize AON engagement; event-related designs with appropriate intertrial intervals; rigorous probe localization with fiducials and digitization; and direct statistical comparisons between execution and observation conditions. The review highlights multimodal opportunities (e.g., concurrent EEG-fNIRS) to fuse spatial and temporal information and supports expanding fNIRS AON research to developmental and clinical populations with paradigms of higher ecological validity (wireless systems, hyperscanning, naturalistic social interactions).

fNIRS is a suitable and promising modality to study the action-observation network, offering improved external validity and motion tolerance relative to fMRI and better spatial specificity than EEG for AON paradigms. Across 14 adult studies, fNIRS revealed stronger, localized execution-related activation in contralateral motor regions, broader bilateral observation effects, and consistent parietal involvement across conditions. The review establishes methodological guidance (bilateral probes, live stimuli, event-related designs, rigorous probe localization, direct condition comparisons) and outlines future directions: whole-head coverage to map all implicated AON regions, multimodal EEG-fNIRS fusion, wireless systems and hyperscanning for naturalistic interaction, and adaptation to infants, children, and neurodevelopmental populations (e.g., ASD) with longitudinal designs. These steps can refine understanding of AON structure, timing, connectivity, and development, and inform rehabilitative and clinical applications.

• Probe placement/reporting: Many studies inadequately measured or reported optode positions relative to fiducials, complicating cortical localization and cross-study comparability.
• Spatial resolution/depth: fNIRS cannot measure subcortical structures and has lower spatial resolution than fMRI; penetration depth is limited (~2–3 cm), restricting measurements to cortex.
• Signal-to-noise and motion: fNIRS has lower SNR than fMRI and is susceptible to instrumental noise, physiological oscillations, hair-related optical coupling issues, and probe motion; requires careful scalp preparation, secure probe fixation, and inclusion of short-separation channels to mitigate superficial noise.
• Paradigm variability: Wide variation in tasks, stimuli (pre-recorded videos vs live), timing, and trial numbers increases heterogeneity; many initial studies lacked both execution and observation conditions, hindering AON inference.
• Analytic inconsistencies: Frequent reliance on condition vs baseline contrasts without direct execution–observation comparisons limits identification of overlapping AON nodes.
• Hemispheric coverage: Predominance of unilateral probes may miss bilateral/lateralization effects; most samples included only right-handed participants, limiting generalizability.
• Ecological validity: Reliance on video stimuli can attenuate AON activation compared to live interactions.