27 healthy experimental subjects (EX; 13 male, 14 female) and 11 healthy controls (HC; 5 male, 6 female) participated in the study. Inclusion criteria for participation in the study included age between 20 and 40 years, right-handedness, no history of neurological or psychiatric diseases, no drug use before the study, native Russian speakers, no alcohol consumption for at least 48 hours before the study, no tobacco or caffeine consumption for at least 2 hours before the start of the study, and no use of psychoactive substances. All subjects underwent a clinical examination and resting-state functional magnetic resonance imaging (MRI).

The experiment involved the administration of a 5-minute calf and foot massage, preceded and followed by a rest period, as shown in Fig. 1. Thus, for each participant we had two 8-minute periods of interest: Rest1 and Rest2. In addition, there was a control group in which subjects only lay down in the scanner for 5 minutes instead of receiving a massage. Two resting-state fMRI sessions were acquired for each subject. Each resting-state fMRI session lasted 480 seconds. The massage was performed by two specialists at the same time to rule out any laterality effect.

The massage stimulation consisted of four sequential movements described below. The sequence of movements was repeated five times from the first to the fourth. The whole procedure took 5 minutes and a stopwatch was used.

(1) Massage movements with the thumbs from the centre of the leg (margo anterior tibiae) upwards about 2–3 cm and then towards the calf muscle on both sides, with the massage movements directed from the foot to the knee. The pressure is moderate. Duration: 15–16 seconds.

(2) Broad circular movements with the palm of the hand in a clockwise direction from the knee to the foot. The pressure is medium to light. Duration: 10–11 seconds.

(3) Friction movements with the palms of both hands from the foot to the knee and back. The movements are quick, with moderate to strong pressure. Duration: 15–16 seconds.

(4) Broad stroking movements with both palms from the knee to the foot and back. The movement is repeated twice with light pressure. Duration: about 18–20 seconds.

We asked participants to rate the pleasantness of the massage using a scale of 1 to 5, where 1 indicated “unpleasant”, 2 was “more unpleasant than pleasant”, 3 represented “neutral”, 4 stood for “more pleasant than unpleasant”, and 5 signified “pleasant”. The results showed no variation in assessments, as all participants rated the procedures as pleasant. After the study, we conducted additional interviews with the participants about their experience of the massage procedures and invited them to suggest any improvements. The majority (23 subjects) indicated that there was nothing to improve, while a few (4 subjects) mentioned that the MRI setup was a limitation.

Based on these findings, we conclude that the social touch procedure, performed in the form of massage, exhibits characteristics associated with an affective type of touch.

Functional and anatomical images were acquired using a 3.0-T Philips Achieva (Koninklijke Philips NV, Amsterdam, The Netherlands) with a 20-channel head coil. Each functional run consisted of 360 T2⁢-weighted echoplanar images, each resting condition—240 slices. The imaging parameters were 2 $\times$ 2 mm in-plane voxel size, covering the entire brain volume in 4 mm slices; interslice gap = 0 mm; repetition time (TR) = 2000 ms; echo time (TE) = 30 ms; and 76 $\times$ 74 matrix, high-resolution T1 image 1 $\times$ 1 $\times$ 1 mm). Participants were instructed to relax and lie still.

The data were processed to examine functional connectivity across the two groups using network-based statistics.

To identify significant differences in global network measures between the control group (HC) and the experimental group (EX), we performed statistical tests for each metric using the t-test. To assess the normality of the distributions of these measures, we applied the Kolmogorov-Smirnov one-sample test to ensure that the data met the normality assumption necessary for the validity of the t-test results.

To identify statistically significant differences in the functional connectivity between the groups, we applied the framework of network-based statistics (NBS) [30]. Specifically, we assessed significance at the p = 0.05 level using the t-test for pairwise comparisons involving the two rest conditions and 50,000 permutations.

We then defined regions of interest (ROIs) based on the results of the group-level statistical analysis using NBS. This was done to reduce the search space in subsequent analyses. We defined the ROIs as the network nodes or brain regions characterized by their inclusion in the maximum number of significantly changing connections according to the NBS results. For these ROIs, we examined the local network metrics and compared them between groups, using the t-test with Bonferroni correction to adjust for the multiplicity of tests due to the number of ROIs considered. This statistical method adjusts the significance threshold to keep the family-wise error rate (FWER) at a desired level, thus reducing the risk of false positives due to the large number of comparisons performed while maintaining statistical rigor.

Statistical analysis of the global network metrics allowed us to identify some important trends, in particular to note that there are no significant between-group differences for all network measures during condition Rest1 (see Table 1). However, for condition Rest2, we see near significant between-group effects on measures of global efficiency, node strength, and clustering coefficient. In addition, for the control group, there is no significant difference between Rest1 and Rest2, while the experimental group shows near significant difference for the global efficiency measure. Thus, global efficiency has near significant effects both in the between-groups comparison in Rest2 and for the experimental group between conditions.

In the Rest1 condition, no significant changes in connectivity were observed between the EX and HC groups. However, in the Rest2 condition, the connectivity in the HC-EX direction showed significant changes in 25 connections (p = 0.038, Fig. 2 and Table 2). In general, the situation is characterized by a limited number of hubs with a moderate number of connections clustered around them. Differences in functional connectivity, particularly around the thalamus and anterior cingulate cortex regions, were found between subjects from the EX group and healthy controls, as shown in the provided Table 2 and Fig. 2. The maximum difference between the groups was found for the Putamen R – anterior cingulate cortex (ACC) pre R connection (0.28), while the highest absolute values for both groups were found for the Caudate L – Temporal Sup R connection (0.48—for the HC group, 0.29—for the EX group). Observably, there is a notable absence of activity in the average connections Temporal Sup R – ACC sub R and Thal VL R – ACC sub R in the EX group, whereas the connection activation is preserved in the HC group.

ROIs were restricted to the five nodes (see Table 3) that participated in the largest number of significant connections previously revealed by the NBS method. We calculated local network metrics for the given nodes and compared them statistically between groups. We found that only the left anterior pulvinar thalamus (Thal PuA L) and right pregenual anterior cingulate cortex (ACC pre R) nodes showed significance for most measures (see Table 3). For betweenness centrality, no node was significant after correction. For eigenvector centrality, the only node found significant was ACC pre R. For the remaining measures—node strength and clustering coefficient—the nodes ACC pre R and Thal PuA L showed significant effects.

Based on the one-sided prevalence of significant connections in the control group, we can assume that this corresponds to the norm and indicates hypoactivity during the experiment in the EX group.

In particular, the anterior pulvinar has connections with the prefrontal cortex, parietal cortex, and other association areas that are critical for complex sensory processing and motor planning [31]. Evidence of abnormal thalamic connectivity in autism spectrum disorders (ASD) and sensory processing disorders suggests that the thalamus may play a role in sensory overreactivity [32, 33]. The mediodorsal magnacellular (MDm) nucleus is also involved in social and cognitive functions, including social cognition and decision-making. It has connections with the prefrontal cortex, which is involved in these processes [34].

The ACC is generally involved in attention and cognitive control. Touch can serve as a sensory cue that attracts attention and may play a role in the integration of this sensory information [35, 36, 37]. The ACC, including the pregenual ACC, is involved in the modulation of pain perception [38]. Touch can have an analgesic effect, in part by activating brain regions that help dampen the perception of pain. The ACC pre is involved in this process because it interacts with other areas of the brain, such as the insula and prefrontal cortex, to regulate emotional and cognitive responses to pain [39, 40]. This area is known to be affected by touch massage [41]. Among its many anatomical connections, the ACC receives nociceptive afferents from the thalamic nuclei and also has connections to the periaqueductal gray and amygdala, while functionally it is involved in specifying the affective content of noxious stimuli and learning to predict and avoid noxious stimuli [19, 42]. The ACC pre is involved in processing the emotional aspects of social interactions, including touch. It helps interpret the emotional meaning of tactile stimuli [43], which increases positive functional connectivity between the cingulate cortex and left anterior supramarginal cortex following mechanical affective touch therapy.

The putamen helps coordinate and execute movements in response to touch, such as reaching out to touch an object or withdrawing from a painful touch. It receives input from the somatosensory cortex, which provides information about touch, and integrates this information with motor plans and intentions. This integration allows for appropriate motor responses to tactile stimuli [44, 45, 46]. The decreased activation in the somatosensory cortices over time may represent stimulus habituation, according to the experimental work [47, 48].

Temporal superior is also involved in social cognition, including the perception and interpretation of others’ actions and emotions. It is particularly important for understanding the intentions and emotions of others based on their touch. For example, the temporal superior can help us interpret a pat on the back, and is known to be involved in predicting the pleasantness of skin stroking [49]. Interestingly, in the same paper [43], there is a reduction in functional connectivity during touch in the ASD subjects. The superior temporal region is a vital part of the mirror neuron system and play a significant role in social cognition, as other research has shown [50, 51].

Although we emphasise the importance of the ACC and pulvinar in this study, many previous studies using fMRI [52, 53, 54, 55] and Near-infrared spectroscopy (NIRS) [56] have highlighted the importance of the orbitofrontal cortex (OFC) and the insula in affective touch. The importance of the OFC has also long been demonstrated in animal electrophysiological studies [57, 58]. The next study showed differentiation in the head of the hippocampus associated with activity in specific limbic brain areas known to be involved in emotional processing [59], which is important in the diagnosis of bipolar disorder. Our findings may have potential in depression treatment trials [60] and other psychological conditions requiring emotion regulation with social touch [61].

The study highlighted the potential of relatively short massage sessions to reduce the activation in brain areas associated with reduced social anxiety and stress. Specifically, the abnormal activity in the discussed anterior pulvinar thalamus and the pregenual anterior cingulate cortex has previously been associated with anxiety-related disorders, including post-traumatic stress disorder (PTSD) and major depressive disorder [62, 63, 64]. The use of massage techniques such as those presented can be successfully used in the rehabilitation and treatment of patients with PTSD and high levels of psychological distress. In addition, the study of resting-state changes based on fMRI data in patients with high levels of psychological distress is the focus of our future research. One of the papers shows that people who received more affectionate touch from their partner during a stressful laboratory task experienced less stress [65].

In addition to its effects on pain and stress, the research has shown that massage therapy is also effective in improving the quality of life of people with multiple sclerosis [66]. In recent years, the use of mechanical affective touch therapy has gained attention as a promising treatment option for controlling symptoms of anxiety disorders [67]. This approach potentially offers a non-pharmacological approach to anxiety management [68]. In line with the attachment theory, the research has shown that social touch experiences can have significant psychological benefits for individuals [69]. These touch experiences can increase feelings of safety and security and improve sleep quality [70], as research has found notable correlations between touch deprivation and self-reported difficulties in initiating and maintaining sleep [71]. Furthermore, study has shown that affection deprivation is significantly associated with sleep disturbance [72].

The relatively small sample size is a limitation of the study that may affect the generalizability of the results. However, the homogeneity of the demographic characteristics of the subjects and the use of adjustments for multiple comparisons during statistical analysis with cluster-based permutation testing are steps taken to mitigate this limitation and improve the validity of the study results.

A potential limitation of the study is the difference in sample size between the experimental and control groups, which could affect the representativeness and generalizability of the results. However, due to practical constraints and limited availability of participants, it was not possible to include a larger sample size in the control group.

The subjects had no previous experience of participating in similar massage experiments. However, it seems reasonable to assume that the subjects have had experience of receiving massage in their lifetime. This factor was not controlled for, and potentially the subjects may have responded differently based on their previous experience.

It is challenging to differentiate the precise influence of social interaction and direct tactile stimulation with its associated affective properties during the massage process. As the subjects were not able to see the masseur in the MRI machine, we believe that the observed effects are mainly due to the tactile influence (touching during the massage). Furthermore, the aim of the current study was to investigate the effects of massage with its potential social aspect.