Introduction

Touch is an essential building block in constructing the foundations of social interaction, attachment, and emotional connection1. In certain neurological disorders, such as autism spectrum disorder, abnormalities in tactile processing and loss of social skills often coexist and are associated with each other2. Studies have observed that tactile deprivation early in life is associated with social deficits in adulthood3,4. Even when visual, auditory, and olfactory communication is present, a lack of tactile contact can significantly increases anxiety levels and alters social skill5. Touch plays an indispensable role in social interaction, as individuals receive crucial information through tactile stimulation during this process, thereby enhancing their understanding of communication. However, the existence of other sensory-mediated communication modalities such as vision, hearing, and smell in the social process, arouses significant difficulties in identifying the role of tactile contact. This may be a fundamental reason why the neural mechanism of social touch is still unclear. Therefore, it is warranted to analyze the tactile and non-tactile contact during social interactions comparatively.

Personal social relationships primarily rely on the brain for processing. When receiving input from social information, specific brain regions exhibit enhanced activity which then enable responses and decision-making6. As a type of sensory input, sensory-related brain regions are critical for perceiving social touch. Previous studies have observed that the primary somatosensory cortex (SI) is activated during tactile contact, while language and eye communication are more involved in hearing and the visual cortex7,8,9. Nevertheless, the role of SI in social interaction remains controversial with some studies suggesting that SI is not directly involved in affective social touch10. In addition, the insular cortex (IC) is a key region in performing various multisensory and emotional processing functions. But this region is also involved in processing other sensory modalities, and whether it responds specifically to social touch has been difficult to determine11,12,13,14. The thalamus is another essential part of sensory transmission and plays a role in sensing social emotions. But there are many subregions in the thalamus, and the specific location that affects social touch is not very clear15,16,17. Other regions associated with higher-order behavior, such as the medial prefrontal cortex (mPFC) and amygdala, also exhibit increased activity during social interaction18,19, the executive functions of such regions are relatively complex, and whether they play a targeted role in touch-mediated social interaction remains to be clarified.

The coordinated activity of brain regions underlies various behavioral modulations. The perception and processing of touch rely on bottom-up cross-regional circuits. Here, SI and IC represent the initial regions within the tactile perception and processing network. These brain regions are responsible for transmitting tactile signals received from the peripheral nervous system to other regions20. For example, the amygdala can receive input from the SI and IC, then further establish functional connections with the mPFC to build pathways for social interaction, emotional response, and regulation21. Other study has shown that pleasant social touch can be conducted along the periaqueductal gray matter of the midbrain to the paraventricular hypothalamus circuit22. However, existing research mainly focuses on limited brain regions. Recently, network-based methods have brought new insights into how the brain generates complex states and behaviors. Established functional connectivity networks, derived from correlations between active regions, have begun to provide a more comprehensive view of brain activity characteristics related to various behaviors and drug effects23,24,25, but no studies have yet applied it to exploring sensory processes in social interactions.

The current study aimed to identify the neural basis of tactile contact in social interactions by comparing alterations in brain activity and neural networks during social interaction in the presence of physical contact (SIPPC) and social interaction in the absence of physical contact (SIAPC). To begin, we assessed the correlation between these two social interactions and tactile sensitivity. Then measured the activation of 100 brain regions during social interactions using c-fos immunohistochemistry. Interregional Pearson correlations were used to identify brain regions that exhibit extensive connectivity to other regions and co-activated networks during SIPPC and SIAPC. Following identified the brain regions that showed specific activity and increased connectivity in SIPPC, we subsequently used a chemogenetic approach to localize the identified brain regions and circuit to validate their role in SIPPC activity (Fig. 1). Our research innovatively applied the methods of whole-brain neuronal activation map construction and brain functional network analysis to explore the differences in neural activities of various social sensory communications. This study established a robust conceptual framework and behavioral paradigm to enhance the understanding of the role of touch in social interactions. Furthermore, the observed differences in neuronal activation and brain network connectivity provide a basis for exploring related mechanisms.

Fig. 1: Procedures for this study.
figure 1

Note: the schematic diagram of the mouse cage in this figure, as well as this element in other parts of the manuscript, is sourced from Scidraw (ID:https://doi.org/10.5281/zenodo.5496326).

Full size image

Results

Tactile sensitivity is associated with SIPPC

To investigate the role of touch during social interactions, two distinct social paradigms were designed: (1) SIPPC-based, where mice directly interacted with an unfamiliar mouse, allowing for physical contact (Fig. 2A); 2) SIAPC-based, where two mice were separated by a partition, enabling them to hear, see, and smell each other without direct physical contact (Fig. 2B). Before this occurred, mice were isolated for 24 h to increase the mice’ desire to socialize26,27. We defined the physical contact (including climbing, face-to-face touching, social grooming, and other forms of physical contact) as tactile contact in SIPPC, and mice within 2 cm of the partition as social approach in SIAPC. In addition, we conducted a series of behavioral assessments to evaluate the tactile sensitivity of the mice, and performed a correlation analysis with the aforementioned tactile contact or social approach time to explore the influence of tactile sensitivity on social interactions.

Fig. 2: Correlation analysis of tactile sensitivity and social duration in SIPPC and SIAPC.
figure 2

A, B Schematic diagram of the SIPPC and SIAPC paradigm. C Analysis of correlation between cotton swab dynamic stimulation response rate and duration of tactile contact in SIPPC. D Analysis of correlation between cotton swab dynamic stimulation response rate and duration of social  approach in SIAPC. E Analysis of correlation between paintbrush dynamic stimulation response rate and duration of tactile contact in SIPPC. F Analysis of correlation between paintbrush dynamic stimulation response rate and duration of social  approach in SIAPC. G Analysis of correlation between whisker deflection rate and duration of tactile contact in SIPPC. H Analysis of correlation between whisker deflection rate and duration of social approach in SIAPC (n = 27–32 mice).

Full size image

Initially, through comparison, we demonstrated that there was no statistical difference in the tactile assessment of mice assigned to both groups (Supplementary Fig. 1), eliminating the potential effects on social interaction arising from varying tactile sensitivities between either group. In addition, the dynamic tactile responses of mice were assessed using a cotton swab/paintbrush that was rapidly brushed across the plantar surface of the hind paw or whiskers, and static tactile responses were measured using von Frey fiber filaments. The following correlation analysis results showed that the average positive reaction rates of the bilateral plantar to cotton swabs (Fig. 2C, r = −0.3530, P = 0.0475) and paintbrushes (Fig. 2E, r = −0.3849, P = 0.0296) were negatively correlated with the duration of tactile contact during SIPPC. Therefore, the higher the response rate to tactile stimulation, the less of tactile contact. However, the correlation between the mice’s withdrawal threshold to von Frey fiber stimulation and duration of tactile contact was not statistically significant (Supplementary Fig. 2A). In addition, we assessed the sensitivity of mouse facial whiskers to stimulation using paintbrush bristles and observed that the whisker deflection rate was negatively correlated with time of tactile contact (Fig. 2G, r = 0.3846, P = 0.0476). Notably, none of the correlations between all tactile assessment results and social approach time during SIAPC were statistically significant (Fig. 2D, F, H and Supplementary Fig. 2B). Thus, the sensitivity to dynamic tactile stimulation in mice correlated with social interactions that involved tactile contact; however, this association was not observed in interactions where tactile contact was absent.

Commonalities and differences in brain activity during SIPPC and SIAPC

To explore whether there are differences in brain activity during SIPPC and SIAPC, immunofluorescence was used to perform a whole-brain screening of c-fos+ cells activated by SIPPC and SIAPC. Thereby, generating whole-brain activity maps for 100 brain regions through comparative analysis with the Allen Mouse Brain Map (Allen Co., USA), and conducted analysis of c-fos data. The c-fos expression of mice single housed for 24 h was used as a baseline reference, with a previous study suggesting that such situations downregulated basal c-fos activation28. To accurately assess the impact of target behaviors on c-fos expression during social interactions, we selected mice that exhibited a higher frequency of tactile contact in SIPPC for c-fos detection. Similarly, the mice included in the detection also had enough time to engage in non-tactile contact during SIAPC. The dataset detailing specific behaviors and their durations is provided in Table S2. The results indicated that, compared to the single-housed (control) group, SIPPC activated a broader range of brain regions, involving 86 regions (Fig. 3B), whereas SIAPC activated 23 regions (Fig. 3C). Subsequently, we conducted a comparative analysis of brain activity during SIPPC and SIAPC, revealing significant differences across 50 regions and all these regions exhibited higher activation levels during SIPPC (Figs. 3A and 4). For detailed descriptions of the differential brain regions among groups, please refer to Table S3.

Fig. 3: Brain regions exhibiting significant activation differences between SIPPC, SIAPC, and the control group.
figure 3

A Difference in the proportion of c-fos-positive cells in the brain between SIPPC and SIAPC. B Difference in the proportion of c-fos-positive cells in the brain between SIPPC and single-housed group. C Difference in the proportion of c-fos-positive cells in the brain between SIAPC and single-housed group. *PFDR < 0.05, **PFDR < 0.01, ***PFDR < 0.001. (n = 5 mice per group, brain regions were divided into major anatomical partitions, as shown on the left side of the figure).

Full size image
Fig. 4: Brain regions exhibiting significant activation differences between SIPPC, SIAPC, and the control group.
figure 4

From left to right are the SIPPC, SIAPC, and single-housed group. Full names represented by abbreviations for all brain areas are shown in Table S1, and pictures of immunofluorescence staining of other brain regions are shown in the Supplementary Fig. 3.

Full size image

To further identify the brain regions specifically activated during SIPPC, SIAPC-activated regions were excluded from the SIPPC-induced regions set. Subsequently, we extracted the regions that exhibited higher activation levels in SIPPC than in SIAPC, generating in a list of 42 brain regions including Cg1, Cg2, PrL, IL, DTT, Pir, DI, AID, AIV, AIP, Cl, pv, Acbsh, AcbC, LSI, LSD, MS, PVA, PT, sm, IAM, PoT, PIL, SPF, Pa, MCLH, BMA, SLEA, La, GrDG, CA3py, SHi, BST, VDB, VTA, MnPo, LPO, DpWh, DpG, InWh, InG and IGL (Fig. 5).

Fig. 5: Venn diagram of the relationship between activated brain areas in different groups.
figure 5

Differences in the coordination of activity between SIPPC and SIAPC.

Full size image

Coordination among multiple brain regions underlies behavioral regulation, and the synchronous activation of different regions often suggests the presence of functional connections between them29,30. To further investigate potential differences in coordinated activities between the brain regions during SIPPC and SIAPC, we calculated the Pearson correlation coefficient (r) between each pair of brain regions in the subjects in each group based on the c-fos measurement values, and compared the differences in r between each brain region and 99 other regions in the two activities separately. After statistical testing, we observed 52 regions with significant differences in r values between the SIPPC and SIAPC. Specifically, in the SIPPC group, 23 regions exhibit a mean r value with other regions that is higher than SIAPC group, whereas 29 regions demonstrate a lower mean r value compared to SIAPC (Fig. 6A, B and Supplementary Fig. 4). Subsequently, the previously screened 42 regions with upregulated activation during SIPPC were intersected with the above 23 regions to obtain a set of 14 brain regions, namely Cg1, Cg2, PrL, IL, AID, AIV, Cl, LSI, LSD, PIL, CA3py, Dpwh, InWh and InG (Fig. 6C). Thus, the activation of these brain areas may be accompanied by the synchronous activation of other areas, which jointly participate in SIPPC activities.

Fig. 6: Differences in coordinated brain activity during SIPPC and SIAPC.
figure 6

A List of certain regions where the mean r value of the SIPPC group is higher than that of the SIAPC group. B List of certain regions where the mean r value of the SIPPC group is lower than that of the SIAPC group. C Venn diagram of the relationship between regions of increased activation and regions of increased coordination in the SIPPC group. *P < 0.05; ** P < 0.01; ***P < 0.001.

Full size image

Brain network topology differences between SIPPC and SIAPC

First, we constructed brain activity networks for SIPPC and SIAPC to characterize the co-activation patterns of all pairs of brain regions present in both groups (Fig. 7A, B). Subsequently, to clarify the differences in brain functional connectivity associated with the two types of social interactions, we extracted the brain regions exhibiting variations in SIPPC and SIAPC connections and constructed respective networks (Fig. 7C, D). Next, we further analyzed the differences in brain activity networks under the two social activity conditions by comparing three network parameters, degree centrality, betweenness centrality, and node efficiency. Among them, degree centrality serves as the most direct metric for assessing node importance in network analysis, a higher node degree indicates greater degree centrality and signifies a more crucial role within the network31. Betweenness centrality assesses a node’s significance by calculating the shortest paths between all pairs of nodes. Specifically, nodes with high betweenness centrality are those frequently traversed in these shortest paths, thereby possessing substantial control and influence within the network32. Node efficiency evaluates the effectiveness of information transfer between different nodes, highly efficient nodes in brain networks usually mean that they play an important role in brain functional connections33. The results suggested that the differences in these three parameters between the SIPPC and SIAPC networks were not statistically significant (Fig. 7E–G). However, when examining each brain region individually, differences in the significance of their roles within the two brain networks emerged.

Fig. 7: Analysis of brain functional networks involved in SIPPC and SIAPC.
figure 7

A Functional connectivity network in the brain during SIPPC. B Functional connectivity network in the brain during SIAPC. C Differential network of SIPPC. D Differential network of SIAPC. E Comparison of the mean differences in brain network degree centrality between the two groups. F Comparison of the mean differences in betweenness centrality between the two groups. G Comparison of the mean differences in node efficiency of the diencephalon network between the two groups. H Brain regions with higher degree centrality in the SIPPC group than in the SIAPC group. I Brain regions with higher betweenness centrality in the SIPPC group than in the SIAPC group. J Brain regions with higher node efficiency in the SIPPC group than in the SIAPC group. K Brain regions with lower degree centrality in the SIPPC group than in the SIAPC group. L Brain regions with lower betweenness centrality of the SIPPC group than in the SIAPC group. M Brain regions with lower node efficiency of the SIPPC group than the SIAPC group. N Venn diagram illustrating the relationship between regions exhibiting enhanced activation and coordination in the SIPPC, alongside key brain regions associated with the SIPPC network. Error bars, standard error of the mean.

Full size image

To this end, we compared the differences in the three parameters of each brain region in the two networks, and listed which brain region parameters were higher in SIPPC (Fig. 7H–J) and which were higher in SIAPC (Fig. 7K–M). Furthermore, the values of the three parameters were combined to define the key regions of the brain network, as these regions exhibited the top five degree centrality and efficiency centrality values, and had betweenness centrality values > 0.1. This definition was used to identify 12 key brain regions in the SIPPC activity network: Rh, IL, Lent, VLG, AIV, LH, PH, PR, VMH, Op, DI, and LSI. This indicates that these regions have established favorable, functional connections with other regions in the SIPPC network.

Therefore, these 12 regions were intersected with the 14 previously screened regions to identify regions where activity and coordination increased and established extensive functional connections with other brain regions during SIPPC. Such activities may promote the elevated activation levels found throughout the community, facilitating collaborative processing of social interactions. The regions that met these criteria include IL, AIV, and LSI (Fig. 7N). Notably, these three regions exhibited lower parameters in the SIAPC brain network, suggesting that they may only exhibit substantial connectivity changes when tactile contact is present during social interactions. Thus, these results suggest that the presence or absence of tactile contact during social interactions affects brain activity and functional connectivity between regions.

Key brain region and circuit are activated differentially during SIPPC and SIAPC

To more precisely observe the relationship between neuronal activity in the brain and differential social interactions, we conducted in vivo calcium imaging in the key brain regions and circuit while the mice were engaged in SIPPC and SIAPC. We expressed GCaMP6m in the IL, AIV, LSI, and AIV neurons projecting to the LSI using adeno-associated virus (AAV) (Fig. 8A–D). Then recorded neuronal calcium activity during tactile contacts in SIPPC and during non-tactile contact in SIAPC. Our findings indicated that, compared to SIAPC, the neurons in the aforementioned three brain regions and the AIV–LSI circuit exhibited significantly stronger neuronal activation during tactile contact in SIPPC (Fig. 8E–L). To further eliminate the potential influence of other touching behavior in SIPPC on neuronal activation, we observed the neuronal activity of mice when they touching the cage wall or engaged in touching oneself, and found that neither of these behaviors elicited significant fluctuations in calcium signals (Supplementary Fig. 5). By analyzing the peak calcium signals induced by different activities, we discovered that the peaks triggered by tactile contact during SIPPC were significantly higher than those observed in SIAPC and other touching behaviors (Fig. 8M–P).

Fig. 8: In vivo fiber-optic calcium imaging records the activation differences in the key brain regions and circuit between SIPPC and SIAPC.
figure 8

A–D Representative images of viral expression in brain regions. E Illustration of calcium imaging recording in IL response to SIPPC (left) and SIAPC (right). F Heatmap showing Ca2+ responses elicited by SIPPC (left) and SIAPC (right) in IL. G Illustration of calcium imaging recording in AIV response to SIPPC (left) and SIAPC (right). H Heatmap showing Ca2+ responses elicited by SIPPC (left) and SIAPC (right) in AIV. I Illustration of calcium imaging recording in LSI response to SIPPC (left) and SIAPC (right). J Heatmap showing Ca2+ responses elicited by SIPPC (left) and SIAPC (right) in LSI. K Illustration of calcium imaging recording in AIV–LSI circuit response to SIPPC (left) and SIAPC (right). L Heatmap showing Ca2+ responses elicited by SIPPC (left) and SIAPC (right) in AIV–LSI circuit. M Differences in calcium signal peaks of IL neurons induced by distinct behaviors. N Differences in calcium signal peaks of AIV neurons induced by distinct behaviors. O Differences in calcium signal peaks of LSI neurons induced by distinct behaviors. P Differences in calcium signal peaks of AIV–LSI circuit induced by distinct behaviors. n = 3 mice per group. Scale bar, 100 μm. *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001. Error bars, standard error of the mean.

Full size image

Key brain regions and circuit inhibition reduce tactile contact behavior during SIPPC

Based on the aforementioned analysis, the SIPPC group exhibits higher activity and connectivity levels in the IL, AIV, and LSI regions than the SIAPC group, indicating that these regions may influence cross-regional network connections during SIPPC, increasing brain-wide activation levels. We hypothesize that this may play a crucial role in regulating tactile contact, this theory warrants further verification. Therefore, we bilaterally injected the rAAV-EF1α-hM4Di-EGFP-WPREs virus into the IL, AIV, or LSI regions of C57BL/6 mice. Following 3 weeks of viral expression, we observed and recorded the time during which the mice injected with viruses in the target brain regions actively had tactile contact with companion mice (of the same gender and with similar age and weight) during SIPPC. And we compared the behaviors of two groups of mice injected with Clozapine N-oxide (CNO) or normal saline intraperitoneally respectively (Fig. 9A). Our findings indicated that after the virus was delivered to the AIV or LSI, compared with saline, the CNO injection significantly diminished the tactile contact (AIV: independent samples t test, t = 4.427, P = 0.0013, Fig. 9B–D, LSI: independent samples t test, t = 3.469, P = 0.0038, Fig. 9F–H) and total social contact (AIV: independent samples t test, t = 3.003, P = 0.0133, Supplementary Fig. 6E, LSI: independent samples t test, t = 3.254, P = 0.0058, Supplementary Fig. 6F) of the mice in the SIPPC group, while having no effect on non-tactile behaviors (AIV: independent samples t test, t = 0.5001, P = 0.6278, Fig. 9E, LSI: independent samples t test, t = 1.466, P = 0.1647, Fig. 9I). This demonstrates that the inhibition of AIV and LSI can influence tactile contact during social interactions. However, no effects of IL inhibition on tactile contact (Mann–Whitney U test, U = 13, P = 0.4848, Supplementary Fig. 6A–C), non-tactile contact (independent samples t test, t = 1.522, P = 0.1589, Supplementary Fig. 6D) and total social contact behaviors were observed (Mann–Whitney U test, U = 11, P = 0.3095, Supplementary Fig. 6H). In addition, our network analysis revealed functional connections between AIV and LSI. To assess the impact of neuronal projections between these two regions on tactile behavior, we bilaterally injected the rAAV-EF1a-DIO-hM4Di-mCherry-WPREs virus into the AIV of mice, and subsequently injected the rAAV-CMV-Cre-hGH pA virus into the LSI. The rAAV-CMV-Cre virus retrogradely enters the neuronal cell body of the AIV along the axon, facilitating the expression of hM4Di, thereby allowing for regulation of the AIV–LSI inhibitory circuit. Initially, we observed the expression of the mCherry signal in the AIV, indicating the presence of neuronal axons projected from AIV bodies to the LSI, thereby confirming the existence of the AIV–LSI circuit. Furthermore, we found that, compared to mice injected with saline, the injection of CNO significantly reduced the tactile contact behavior (independent samples t test, t = 2.563, P = 0.0249, Fig. 9J–L) of mice in the SIPPC, and the total social interaction time was also markedly decreased (Mann–Whitney U test, U = 6, P = 0.0175, Supplementary Fig. 6G); however, there was no effect on non-tactile behavior (Mann–Whitney U test, U = 19, P = 0.5350, Fig. 9M). For detailed descriptions of the differential brain regions among groups, please refer to Tables S4–7. This suggests that inhibiting the AIV–LSI circuit specifically diminishes tactile exposure during social interactions.

Fig. 9: Chemogenetic inhibition of AIV/LSI or AIV–LSI circuit reduces the tactile contact behaviors during SIPPC.
figure 9

A Experimental protocols for evaluating the impact of inhibiting AIV/LSI brain regions or the AIV–LSI circuit on social behavior in SIPPC. B Schematic of bilateral virus infection in AIV. C Representative images of viral placements in AIV. D Period of tactile contact during SIPPC. E Period of non-tactile contact during SIPPC (n = 6 mice per group). F Schematic of bilateral virus infection in LSI. G Representative images of viral placements in LSI. H Period of tactile contact during SIPPC. I Period of non-tactile contact during SIPPC (n = 8 mice per group). J Schematic of bilateral virus infection in AIV/LSI. K Representative images of viral placements in AIV. L Period of tactile contact during SIPPC. M Period of non-tactile contact during SIPPC (n = 7 mice per group). Scale Bar, 100 μm. Error bars, standard error of the mean.

Full size image

Discussion

In this study, we confirmed that tactile perception is exclusively associated with tactile contact in social interaction. And then conducted a comparative analysis to examine the effects of two forms of social interaction, SIPPC and SIAPC, on c-fos expression across various brain regions. Our findings indicated that SIPPC activated a broader range of regions compared to SIAPC and exhibited higher activation levels. Further investigation into inter-brain coordinated activity revealed distinct differences between the functional networks engaged during SIPPC and SIAPC. In addition, we determined through screening that the activities of three regions, IL, AIV and LSI, play a critical role during SIPPC and found that there is a functional connection to AIV and LSI. GI-DREADD-mediated inhibition of the AIV, LSI brain regions and AIV–LSI circuit reduced tactile contact during SIPPC. This validates the results of our c-fos data analysis, and suggests that the AIV, LSI brain regions and AIV–LSI circuit may play a crucial role in regulating tactile contact.

Social interactions rely on the interpretation of emotional information from multiple sensory modalities34, which makes it more challenging to comprehensively understand the relationship between tactile perception and social behavior. Herein, the designed behavioral paradigms solve this problem to some extent. The subsequent correlation analysis observed that animals with higher tactile sensitivity reduced tactile contact behaviors during social interaction; however, this did not affect other interactive behaviors. This phenomenon has also been reported in previous studies, as individuals with tactile hypersensitivity were discovered to be more resistant to tactile contact due to them receiving less pleasure from perceiving touch than the general population35. This aversion may be sufficiently extreme to cause them to perceive touching as painful36,37. Thus, the study of social behaviors that depend on different senses is necessary for the elucidation of the mechanism whereby touch affects social interaction.

Utilizing c-fos immunofluorescence staining, widespread neuronal activation during SIPPC was observed, even in regions previously identified as associated with sensation, emotion, and reward, such as the IC (DI, AIV, AID, AIP), VTA, nucleus accumbens (Acbsh, AcbC), mPFC (Cg1, Cg2, IL, PrL)38,39,40. Notably, these regions were not significantly activated during SIAPC, indicating a specific role of tactile contact. Among these brain regions, mPFC is more closely related to tactile contact. Previous functional magnetic resonance imaging studies have shown that mPFC is highly activated during pleasant social interaction41,42, and is involved in social decision-making and cognition43. Both the thalamus and IC are brain regions closely associated with tactile processing, with the IC being more specifically related to emotional signals44,45. Gentle, dynamic touch is preferentially encoded by a unique neural pathway involving C-tactile (CT) afferents46, with the IC serving as a key region for processing CT signals47,48. While it was previously believed that the IC also processed other sensory signals, our findings demonstrate its heightened relevance in sensing tactile stimuli during social interactions. Notably, the activation of reward-related brain regions, such as the VTA and nucleus accumbens49,50, in response to SIPPC suggests that the reward circuit may play a role in modulating the emotional experience associated with pleasant social touch.

In addition, we observed that the activation of regions including the LS, BST, and MS/VDB were pronounced in the SIPPC group than in the SIAPC group. While this is the first study to report on the relationship between these regions and social touch. However, previous studies implied that the LS and BST are associated with reward behaviors and social interactions51,52. Such information emphasises the role of reward function in social touch. Another study observed that the Gabrb3 gene deletion in BST is the main reason underlying tactile hypersensitivity in autistic mice model53, which verifies the role of BST in tactile perception. Furthermore, oxytocin receptors exhibit high expression levels in LS and BST54,55,56, and we speculate that this may be the mechanism whereby sensing and social touch processing are regulated. Currently, there is insufficient research on the role of MS/VDB in touch and social behaviors. However, these regions are considered the most abundant with basal forebrain cholinergic neurons. Okada et al. observed that mice lacking cholinergic neuron cell populations in the basal forebrain exhibited impaired social interaction or social recognition memory57. Although the role of cholinergic function in social behavior is poorly understood, our results provide reference for research in this field.

The number of activation regions was significantly reduced during SIAPC, with a relatively low activation level, suggesting that tactile contact plays a crucial role in social interaction. Results revealed that activated during SIAPC reveals significant involvement of the olfactory and auditory cortices, including areas such as PRh, Ect, and AuV, along with other brain regions associated with primary sensation, such as the thalamus58,59,60. This suggests that the absence of tactile contact aids in the reduction of activity in brain areas related to emotional and social behaviors. Other regions exhibiting significant activation include DEn, BLA, BMP, and CA1py. Among these, DEn primarily forms a neural circuit with the auditory cortex61,62. Notably, the BLA, BMP, and CA1py regions were activated during SIAPC. Furthermore, c-fos expression was observed in the amygdala during SIPPC, with only the BMA exhibiting a slightly elevated level, whereas the other regions exhibited reduced c-fos expression. Such activity occurs during the SIPPC-induced extensive activation of higher-order brain regions associated with emotions, suggesting that the amygdala may not play a prominent role in the social behavior associated with tactile contact, but rather is involved in the socialization processes linked to other senses. Furthermore, SIPPC impact on hippocampal c-fos expression was mainly observed in GrDG and CA3py, highlighting the differences in activity within various areas of the same brain region during social interactions.

Brain function relies on the coordinated activity of multiple interconnected brain regions63. In this study, we further elucidated the differences in brain coordination during SIPPC and SIAPC by comparing the correlation of c-fos expression among various brain regions. Initially, we examined the distribution of the Pearson r value per brain region in relation to others, identifying those regions with higher mean r values in SIPPC or SIAPC, thereby demonstrating the establishment of stronger functional connections during these activities. Therefore, the trends in differences of brain coordination generally mimic those in activation levels. Specifically, regions exhibiting greater neuronal activation tend to demonstrate stronger coordination. In addition, functional network analysis further evaluated system-level interactions between various brain regions during SIPPC and SIAPC, and parameters such as centrality measures were calculated to evaluate brain network characteristics. Thus, the Rh and LH regions have the highest degree and betweenness centrality, respectively, highlighting their role as the primary connection hubs in the complete network. Rh is located in the ventral midline of the thalamus and exhibits dense interconnections with the anterior mPFC and hippocampus, thereby promoting strong bidirectional effects on the flow of information between these structures. This coordination may contribute to Rh-related behavioral functions, which we hypothesize are essential in engaging in social interaction64,65,66,67. In addition, LH is a key part of the hypothalamus, which was previously believed to be related to eating behavior. Recent studies have observed that the LH can serve as a relay station connecting the forebrain and midbrain circuits. Compared with other parts of the hypothalamus, the LH has stronger efferent and afferent connections with multiple forebrain areas and can receive signals from the forebrain. Information about sensory cues, previous experiences, and brain states are projected to the midbrain reward areas, enabling the rapid coordination of cognitive and reward processing circuits to control adaptive social contact behavior68,69,70. Herein, the higher betweenness centrality value of the LH region in the SIPPC network further emphasises its mediating role in social interactions with tactile contact.

Notably, we performed three progressive screenings during the analysis. Through the first screening, we identified regions that were specifically induced by SIPPC (Fig. 5), then we intersected them with those brain regions that exhibited increased levels of co-activation with other regions (Fig. 6C). From the second screening results, brain regions that also belong to key nodes of the SIPPC network were extracted (Fig. 7N). Finally, we observed that the IL, AIV, and LSI regions corresponded with the above three conditions simultaneously. Further results verified that the inhibition of the AIV and LSI significantly reduced tactile contact behavior, and confirmed the presence of neuronal projections between the AIV and LSI. Chemogenetic inhibition of the AIV–LSI circuit substantially diminished tactile contact during social interactions, suggesting that the AIV may modulate social touch behavior through its projections to the LSI. Although the functional circuitry between the IC and LS remains to be fully elucidated, previous studies have indicated that the AIV is a subregion of the IC, an integrative center within the brain that receives sensory input from both external and internal sources at the thalamic and cortical levels71,72 and subsequently relays it to areas associated with motivation and reward73,74. The LSI, a subregion of the lateral septal nucleus, plays a crucial role in regulating emotion, reward, and social behavior, and modulates various neuronal activities by integrating signals from different neocortical and neuromodulatory systems, influencing emotional states and behavioral responses75,76,77,78,79. Therefore, we speculate that the effect of the AIV–LSI circuit on social touch behavior may be linked to its role in the emotional processing of tactile contact in mice during social interaction. An interesting result is that, although both our in vitro and in vivo experiments indicate that several connections in the infralimbic (IL) brain region are significantly activated during SIPPC activities, inhibiting the IL brain region does not exhibit a significant impact on tactile contact behavior. We speculate that this may suggest that the IL region regulates tactile contact behavior through mediation or redundancy effects within the brain network. The specific mechanisms underlying this phenomenon warrant further investigation in future research80,81.

This study contributes to the discourse on the relationship between social interaction and touch. We first established that tactile sensitivity is directly correlated with tactile contact. In addition, our findings indicate that the presence or absence of tactile contact during social interactions elicits distinct activation patterns in the brain, which in turn lead to variations in functional network connections. Importantly, we identified and validated key brain regions and neural circuits primarily responsible for regulating tactile contact behavior. These results provide a solid foundation and direction for further exploration of the neural mechanisms underlying social touch. However, this study does have certain limitations. One limitation is that, due to the higher proportion of male patients in neurodevelopmental disorders characterized by social deficits, we exclusively used male mice as experimental subjects, and there is not a large sample size. Thus, future research should increase the number of mice and further explore the activity characteristics of female brains during social interactions. In addition, we utilized isolated brain slices to examine c-fos expression, aiming to elucidate the effects of social contact behavior on the activation and connectivity of brain neurons. Compared to in vivo imaging experiments, this method exhibits less synchronization in correlating behavior with neuronal activity. Although we validated the efficacy of this approach by observing key brain regions through fiber-optic calcium imaging, the extensive number of brain regions analyzed presents significant challenges for individual verification. Future research will employ more advanced experimental techniques to facilitate in-depth exploration across a broader array of brain regions.

Methods

Animals

All adult male C57BL/6J mice, aged 8–12 weeks and weighing between 20–26 g, were acquired from Liaoning Changsheng Biotechnology Co., Ltd. The mice were housed in an environment maintained at a temperature of 23 ± 2°C and a humidity level of 50 ± 10%, with a 12-h light–dark cycle and unrestricted access to food and water. The mice were allowed 1 week to acclimatize to their new environment prior to the commencement of the experimental protocol. All experimental protocols adhered to the guidelines outlined in the Guide for the Care and Use of Laboratory Animals published by the National Institutes of Health and received approval from the Animal Experiment Committee of the Graduate School of Medicine at Harbin Medical University (Approval number:HMUIRB2023039).

Social interaction paradigms

SIPPC

One day before contact activity was commenced, a pair of mice (the test and companion mice) were isolated in a single cage for 24 h. At the onset of the experiment, both mice were placed in a new cage (20 cm × 15 cm × 12 cm) simultaneously and allowed to interact freely for 20 min. Observe and record the cumulative time during which the test mouse actively make tactile contact with the companion mouse. Recorded behaviors included climbing, face-to-face touching, social grooming, and other forms of physical contact.

SIAPC

The test mouse and companion mouse were each isolated in a single cage for 24 h before engaging in a 20 min social interaction., and then simultaneously placed into a new cage that featured a transparent partition (20 cm × 15 cm × 12 cm). Circular holes measuring 0.6 cm in diameter were drilled into the partition in an 11 by 14 configuration to facilitate the passage of sound and scent. Here, the mice can smell, hear, and see each other, but no direct physical contact is possible. Social approaching behavior when it was within 2 cm of the partition includes mice approaching the partition to explore the head, face, nose or other body parts of the companion mouse. The cumulative duration of social approaching time was observed and recorded.

Tactile assessment

Von Frey test

Tactile thresholds of the hind paws of the mice were assessed using von Frey filaments ranging from 0.008 to 2 g. After allowing the mice to fully adapt to their environment, the filaments were used to stimulate the soles of the hind paws, progressing from low to high intensity. Each mouse received 10 stimuli of the same filament intensity, and the withdrawal threshold was defined as the first filament that elicited a response rate exceeding 50%.

Dynamic brush test

Mice were placed on a metal mesh test bench and allowed to acclimatize for 30 min. Following acclimatization, once the mice were in a calm state, a soft brush was used to gently brush the right and left plantar ankles of the mice towards the plantar toes while avoiding direct contact with the plantar feet. During the test, each mouse was brushed ten times, with an interval of >2 min between each brushing. The percentage of mice that either walked away or lifted their feet during the ten trials was recorded as the outcome of the brush test.

Cotton swab test

The mice were placed on a metal mesh test bench and allowed to acclimatise for 30 min. Subsequently, when the mice were in a calm state, cotton swabs (expanded to thrice their original size) were used to gently stroke the mice from the ankle to the toe of both the right and left plantar surfaces while avoiding direct contact with the plantar surface. During the test, each mouse was gently stroked ten times, with intervals of >2 min between each application. This procedure was repeated ten times, and the percentage of positive responses was recorded to indicate the degree of paw withdrawal.

Whisker deflection

Each mouse was individually placed in a cylindrical mesh chamber and allowed to acclimatise for 5-min to minimize stress. Subsequently, a soft-bristled brush was used to gently stroke the left and right whiskers of the mice while avoiding direct contact with their faces. Each mouse underwent a total of 10 brushing sessions, with intervals exceeding 2 min between each session. The behavioral responses recorded during these tests included the forceful stroking of the forepaw against the face, head shaking, head deflection, and retraction of the head from the stimulus. These responses were quantified as the outcomes of this test.

Immunohistochemistry

In all, 90 min after the end of social interaction, mice were anaesthetized, cardiac perfusion was performed, and their brains were collected. The brains were fixed in 4% paraformaldehyde at 4 °C and subsequently transferred to a 30% sucrose solution. Using a freezing microtome, 40-μm-thick sections were prepared. To assess c-fos expression, coronal brain sections were incubated with a blocking solution (3% normal goat serum, 0.3% Triton X-100) for 1 h at 24 °C. Then, the sections were treated with anti-c-fos primary antibody (1:1000, NBP2-50037, NOVUS, USA) and incubated overnight. After washing, the sections were incubated with anti-mouse Alexa-488 secondary antibody (1:1000, 4408, Cell Signaling, USA) for 1 h at room temperature. The sections were washed again and immediately mounted with an anti-fluorescence quencher containing DAPI. Image scanning was conducted using a digital slide scanner (PANNORAMIC MIDI II), and c-fos expression in various brain regions was quantified using ImageJ software.

Network construction

Generate the SIPPC and SIAPC networks by considering correlations with Pearson r ≥ 0.80 (one-sided significance level of 0.05). Network construction was performed in the Python package networkX (https://github.com/networkx/networkx; version 1.11). Network centrality was evaluated using degree centrality, betweenness centrality, and eigenvector centrality measures to identify potential central regions. The nodes in the network represent brain regions, the correlations retained post-threshold segmentation are used as edges, and the r values are used as edge weights. Degree centrality represents the total number of edges connected to a specific node. The node efficiency of each node is calculated as the average reciprocal of the shortest path lengthwise to all other nodes in the network. Global efficiency refers to the average efficiency of all nodes in the network.

Stereotactic surgeries

At the time of virus injection, mice were placed on a brain stereotaxic apparatus (RWD Life Technologies Co., Ltd., China). Anesthesia was induced using 2% isoflurane in oxygen and maintained with 1% isoflurane in oxygen. A glass microelectrode connected to an infusion pump was utilized to deliver 300 nl of virus to the target brain region at a rate of 50 nl/min. Following the completion of the injection, the glass microelectrode was left in place for 10 min. When using calcium imaging to observe the activity of different brain region, the rAAV-EF1α-GCaMp6m-WPRE-hGH pA were injected unilateral into the IL/AIV/LSI regions of C57BL/6 J mice (IL: AP 1.42 mm; ML ± 0.5 mm; DV −3.0 mm; AIV: AP 1.6 mm; ML ± 2.75 mm; DV −3.8 mm; LSI: AP 0.5 mm; ML ± 0.4 mm; DV −3.6 mm). When using calcium imaging to observe the activity of the AIV–LSI circuit, rAAV(Retro)-CMV-cre-hGH-pA were injected unilaterally into the LSI, and the rAAV-EF1α-DIO-GCaMp6m-WPRE-hGH pA were injected unilaterally into the AIV of C57BL/6J mice. For chemogenetic inhibition of the neural activity in different brain regions, the rAAV-EF1α-hM4Di-mCherry-WPREs were injected bilaterally into the IL/AIV/LSI regions of C57BL/6 J mice. For chemogenetic inhibition of the activity of the AIV–LSI circuit, rAAV-CMV-Cre-hGH pA were injected bilaterally into the LSI and rAAV-EF1a-DIO-hM4Di-mCherry-WPREs were injected bilaterally into the AIV of C57BL/6J mice.

Optical fiber calcium imaging

Two weeks following the virus injection, an optical fiber is implanted into the targeted brain regions (IL/AIV/LSI). After allowing the mice to recover for one week, signals are recorded using the optical fiber photometry system (Thinker Tech, Nanjing Bioscience, Inc.). The mice are placed in either the SIPPC or SIAPC scenarios, where calcium signals are recorded before and after the occurrence of specific target behaviors: tactile contact in SIPPC, and social interaction, wall touching, and self-touching in SIAPC. The transient value of the calcium signal (ΔF/F) is calculated using the formula (F-F0)/F0, where F0 represents the median Ca2+ transient from 2 s prior to the onset of the behavior. Subsequently, the ΔF/F values for all behaviors are averaged.

In vivo pharmacological approach

In all, 30 min prior to the behavioral test, CNO at a concentration of 0.33 mg/ml was administered via injection into the abdominal cavity of the mice at a dosage of 0.2 ml per 20 g of body weight. Mice in the control group received an equivalent volume of 0.9% normal saline.

c-fos analysis process

Before social interaction, both the test mice and companion mice were individually housed in cages for 24 h to enhance their social motivation. Subsequently, mice in the SIPPC group engaged in free social interaction with companion mice in a normal cage without a partition for 20 min, while mice in the SIAPC group participated in social interaction without tactile contact with companion mice in a partitioned cage for the same duration. Control group mice remained alone in an empty cage during this process to minimize baseline c-fos expression. After 90 min, the whole brains of the mice from all three groups were extracted, and frozen sections were prepared for c-fos detection. Based on “The Mouse Brain in Stereotaxic Coordinates”, we quantified the anatomical spacing between adjacent coronal planes using Bregma as the rostrocaudal coordinate origin. Analysis demonstrated that consecutive atlas levels exhibited an average section thickness of 80–160 μm, allowing the acquisition of 2–4 serial 40-μm sections per anatomical plane. We obtained continuous whole-brain sections corresponding to atlas numbers 14–60, which range from 2.1 mm to −3.52 mm from Bregma. From these sections, we selected those that best matched the atlas features for subsequent immunofluorescence staining. Utilize ImageJ to accurately align fluorescence-stained sections with the corresponding brain atlas, delineating the contours of each brain region. Calculate the area of each brain region and quantify the number of positive cells, recording the number of positive cells per unit area as the c-fos activity level for that specific brain region.

Statistics and reproducibility

All statistical tests were conducted using Python 3 (libraries: matplotlib and seaborn), GraphPad Prism 10.0, or SPSS version 26.0. Initially, Pearson correlation analysis was performed to investigate the relationship between the results of the tactile perception tests and the duration of social interaction. Furthermore, one-way ANOVA was utilized to evaluate the differences in c-fos expression levels across multiple groups, followed by Tukey’s HSD post hoc tests to identify specific differences between each pair of groups. Multiple correction was applied using the Benjamini–Yekutieli FDR correction. For other inter-group comparisons, the Shapiro–Wilk test is initially employed to evaluate the normality of the data. The independent sample t test or one-way analysis of variance is subsequently utilized to analyze differences between two or more groups of data that conform to a normal distribution. Conversely, for data that do not adhere to normal distribution, the Mann–Whitney U test or the Kruskal–Wallis test is applied.

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.