A rat model of volitional mutual social interactions

Abstract

Social interactions are essential for building societies and fostering cooperation among individuals. These behaviors are governed by complex norms and signaling mechanisms promoting mutual engagement. While animal models are often used to study social behaviors, they typically focus on one individual, overlooking the role and motivation of an otherwise passive social partner. Here, we developed a model where resident and partner rats voluntarily engage in mutual social interactions. In this model, the resident initiates interaction by pressing a lever to activate cues for the partner, who responds by pressing an additional lever, leading to social interaction. To test motivation for mutual social interaction, we increased the effort required for both residents and partners either concurrently or independently. We further investigated the mechanisms underlying these interactions by manipulating the norepinephrine system both systemically and centrally during mutual social interactions. Both male and female paired rats consistently demonstrate mutual motivation to engage in social interactions, regardless of their roles. The rats effectively coordinate their actions, showing low latency and high engagement frequency even as effort demands increase. The average social score analysis identified a significant proportion of highly motivated social pairs. Manipulating the norepinephrine system selectively disrupted the distribution of highly motivated social pairs, emphasizing its role in regulating social interactions. Ablating norepinephrine terminals had no impact on motivation for food rewards, further confirming that central norepinephrine manipulation specifically affects mutual social interactions. Our findings provide insight into the fundamental behavioral and neurobiological mechanisms underlying sociability and complex social structures in rodents.

Introduction

Social interactions are fundamental to our daily lives, shaping our identities, relationships, and societies [1]. These interactions are complex and multifaceted, involving both verbal and nonverbal cues, cultural norms, and individual differences [2, 3]. Typically, social interactions involve a give-and-take dynamic, where individuals exchange resources or mutual behaviors [4, 5]. Through such interactions, individuals can form alliances, build trust, and resolve conflicts, benefiting both themselves and their social groups [6, 7]. Understanding the behavioral and neurobiological mechanisms behind motivated social interactions is crucial for fostering healthy relationships and addressing societal challenges [8].

The study of animal social behavior, both in the wild and in laboratory settings, seeks to uncover the mechanisms driving these interactions [9, 10]. While human social behaviors raise questions about the conservation of mutual behaviors across species [11], exploring such behaviors requires an evolutionary perspective [12], recognizing that animals often weigh immediate and long-term outcomes [13]. Although natural selection typically disfavors behaviors that reduce individual benefits, social cooperation can emerge when the benefits of interaction outweigh altruistic costs [14, 15]. Mutual social interactions are observed across diverse species, including vampire bats [16], bumblebees [17], rodents [18], parrots [19], and non-human primates [20]. In wild-type Norway rats, cooperation is influenced by the quality of prior assistance and experiences of helping others [21, 22]. By integrating anthropology, psychology, and evolutionary biology, social neuroscience is advancing understanding of the cognitive and social dynamics shaping interactions across species.

Despite decades of research on social behavior, defining and measuring a partner’s social motivation or willingness to engage in mutual interaction remains elusive. A significant limitation of traditional operant social interaction rodent models in experimental settings [23–30] is that interactions are usually controlled either by the experimenter or by a designated experimental resident animal, leaving the partner to passively receive interactions – a setup that does not allow for an assessment of the partner’s willingness to interact socially. To address this gap, we developed a behavioral model in which a resident rat presses a lever to activate social cues for a partner. If motivated, the partner rat responds by pressing another lever, initiating mutual social interaction. Both male and female single-housed rats consistently demonstrated motivation to engage, exhibiting low latency and high interaction frequency, even as effort demands increased. An average social score revealed that a significant proportion of pairs displayed heightened motivation for voluntary mutual interactions.

To investigate the neurobiological mechanisms underlying this motivation, we focused on norepinephrine. This is based on previous research demonstrating that catecholamine manipulation influences social play and other motivated behaviors [31, 32], as well as norepinephrine’s role as a key catecholamine neurotransmitter involved in assigning motivational value to sensory information [33]. While its role in motivation relative to dopamine remains debated, some suggest that catecholamines have dissociable yet overlapping functions [26]. Nonetheless, norepinephrine plays a critical role in social behavior, influencing social decision-making in intergroup relations and moral judgments in humans [34]. In rodents, norepinephrine facilitates social discrimination between novel and familiar juveniles, with depletion impairing this ability [35, 36]. Pharmacological studies further demonstrate its role in regulating social play [37]. Based on this evidence, we hypothesized that norepinephrine modulates motivation for mutual social interactions. Using our behavioral model, we found that systemic pharmacological interventions or central norepinephrine lesions selectively disrupted motivation for social interactions without affecting food reward motivation. These findings highlight norepinephrine’s role in social motivation and provide a foundation for further research into the neural mechanisms underlying social behavior.

Materials and methods

Subjects

We used young adult male and female Sprague-Dawley rats Charles River, total n = 228 rats [125 “Resident” (61 females and 64 males) and 97 “Partners” (48 females and 49 males]), weighing 150–225 g upon arrival. We housed rats in same-sex pairs, matching them by body weight and sex, for two weeks prior to the experiments. Afterward, we individually housed them for the remainder of the study to maintain high motivation for social interaction [27, 38]. Rats were randomly assigned to either “Resident” or “Partner” groups, and their roles were maintained throughout the experiment. The rats were kept on a reverse 12-h light/dark cycle (lights off at 9:00 am) with free access to standard laboratory chow and water. Three pairs were excluded due to complications with the norepinephrine lesion. All procedures adhered to the guidelines in the Guide for the Care and Use of Laboratory Animals and were approved by the UMB School of Medicine Animal Care and Use Committee.

Mutual social interactions chamber

We trained the rats for volitional mutual social interactions and for food self-administration using custom-made mutual social interaction self-administration chambers (Fig. S1). We modified a standard Med Associates social self-administration chamber [27, 30, 38–44] by adding a left panel to the partner chamber. This panel included a discriminative stimulus (white houselight) indicating the availability of the partner-social-paired lever (retractable) and a conditioned stimulus (tone speaker). The left panel of the resident chamber featured a discriminative stimulus (red light) signaling the insertion and availability of a food-paired active lever (retractable), along with a nearby pellet dispenser and a stationary inactive lever. The right panel of the resident chamber included a discriminative stimulus (white houselight) indicating the insertion and availability of the resident-social-paired active lever (retractable).

Procedures

Mutual social self-administration

The session began with the illumination of the resident-social-paired white houselight, followed by the insertion of the resident-social-paired active lever after 10 seconds. The houselight remained on throughout the session as a discriminative stimulus. Successful presses on the resident-social-paired lever caused its retraction and activated the white houselight in the partner chamber, signaling the availability of the partner-social-paired lever after 10 seconds. Pressing the partner-social-paired lever resulted in its retraction, a 20-second tone cue, and the opening of the guillotine door for mutual interaction. Paired rats interacted through a perforated screen for 60 seconds, after which the door closed, and they could restart the interaction. Each pair could receive one reward every two minutes. Rats were trained in daily 60-minute sessions with varying conditions: Fixed-ratio (FR) 1 and a 20-second timeout in Experiment (Exp.) 1 (6 sessions), increased FR requirements in Exp. 2–4 (9 sessions), and progressively increased effort for the partner in Exp. 3–4. Data collected included social rewards achieved, lever presses by both rats, latency to social interaction, and the frequency of mutual social interactions.

Progressive ratio

During the progressive ratio sessions (Exp. 3), we increased the ratio of responses per social reward (only for the partner-paired-lever) according to the following sequence: 2, 4, 6, 9, 12, 15, 20, 25, 32, 40, 50, 62, 77, 95, 118, etc.) over three sessions. The final completed response ratio represents the “breaking point” value.

Food self-administration

Rats were trained to lever press for food in two daily 2-h sessions (Exp. 5; 3 to 9 sessions), separated by 10 minutes, under an FR1-4, 20-second timeout reinforcement schedule. This schedule delivered one 45-mg palatable food pellet (TestDiet, Catalog # 1811155). Before the formal operant training sessions, rats underwent 20-minute magazine training, during which five pellets were delivered non-contingently every 5 min. Each session began with the white houselight, followed by the insertion of the food-paired active lever after 10 seconds. The houselight remained on throughout as a discriminative stimulus. At the session’s end, the houselight turned off, and the lever was retracted. Data collected included the number of food rewards and lever presses (active and inactive).

Drugs

We obtained Prazosin HCL (Cat. # 0623/100), RX821002 (Cat. # 1324/10), Clonidine HCL (Cat. # 0690/100), SCH39166 hydrobromide (Cat. # 2299/10), Raclopride (Cat. # 1810/10), and Ketanserin tartrate (Cat. # 0908/50) from Tocris Bioscience and LY341495 (Cat. # Hy-70059) from MedChemExpress. Before test sessions, all drugs were dissolved in sterile saline, except for Prazosin and RX821001, which were dissolved in 10% DMSO and sterile saline. We injected the rats with the compound 30 minutes prior to the start of the test session. We used a within-subject Latin-square design for all compounds injecting systemically either intraperitoneally (i.p) or subcutaneously (s.c) with varying doses: Prazosin (0, 1, 2 mg/kg – i.p.), RX821002 (0, 0.1, 0.5 mg/kg – i.p.), Clonidine (0, 0.02, 0.04 mg/kg – i.p.), SCH39166 (0, 0.1, 0.2 mg/kg – s.c.), Raclopride (0, 0.05, 0.1 mg/kg – s.c.), Ketanserin (0–2 mg/kg – i.p.), and LY341495 (0–5 mg/kg – i.p.), based on previous literature [45–51]. We conducted separate test sessions for resident and partner rats (Exp. 4–5). Utilizing this protocol, we were able to ensure that each resident received two doses of pharmacological treatment compared a vehicle while also controlling for cumulative effects of treatments. During resident testing, we injected compounds into the resident only, with the effort requirement increased to FR4 (while maintaining FR1 for the partner). Conversely, during partner testing, we injected compounds into the partner only, with the effort requirement increased to FR4 (while maintaining FR1 for the resident). During testing for food reward, we maintained the required effort to FR4 for one pellet. We obtained N-(2-chloroethyl)-N-ethyl-2-bromobenzylamine hydrochloride (DSP-4, Cat. # 2958) from Tocris Bioscience. For DSP-4, we used a between-subject design and balanced the groups based on their performance on the mutual social interaction task (Exp. 4–5). We injected the rats once (0, 50 mg/kg – i.p.) and then tested for mutual social self-administration at increasing FR effort requirements (3 sessions at FR1, 3 sessions at FR2, and 3 sessions at FR4) for both resident and partner rats over two weeks, which is the reported time for DSP-4 to exert its neurotoxic effects [52, 53].

Immunohistochemistry

Immediately after the behavioral tests, we anesthetized the rats with isoflurane and perfused them transcardially with ~100 mL of 0.1 M phosphate-buffered saline (pH 7.4) (PBS) followed by ~400 mL of 4% paraformaldehyde (PFA) in PBS. We removed the brains and post-fixed them in 4% PFA for 2 h before transferring them to 30% sucrose in PBS for 48-h at 4°C. We froze the brains in dry ice and stored them at -80°C. We cut coronal sections (40 µm) using a Leica cryostat. We collected the tissues in cryoprotectant (20% glycerol and 2% DMSO in 0.1 M PBS, pH 7.4) and stored them at −80°C until further processing.

TH and NET double-labeling

We selected a series of brain sections representing approximately one in every four sections throughout the entire brain. We took sections at different bregma levels, encompassing cortical, striatal, thalamic, amygdala, and other midbrain regions, as well as regions of the cerebellum. We collected the sections from bregma levels ranging from +4.6 to -10.6 [54] for each rat. We used immunofluorescence to double-label Tyrosine Hydroxylase (TH) with Norepinephrine Transporter (NET). We rinsed free-floating sections in PBS (3×10 min), incubated for 1 h in 10% normal horse serum (NHS) in 0.5% PBS-Tx, and incubated the sections for 48 h at 4°C with chicken anti-TH primary antibody (1:1000, Biorbyt, orb22938, RRID: AB_10928487) and mouse anti-NET primary antibody (1:1000, Thermo Fisher, CL3063, RRID: AB_2637262) in 4% BSA in 0.3% PBS-Tx. We rinsed the sections in PBS (3×10 min) and incubated them for 4 h with goat anti-chicken Alexa Fluor 488 (1:1000, Thermo Fisher, A-11039, RRID: AB_2534096), donkey anti-mouse Alexa Fluor 594 (1:1000, Thermo Fisher, A-21203, RRID: AB_141633) in 2% NHS in 0.5% PBS-Tx. We rinsed the sections three times in PBS (3×10 min) and mounted them onto gelatin-coated glass slides, air-dried, and coverslipped the sections with mowiol + DAPI (Millipore). To collect the images, we used an Olympus VS200 Slide Scanner with the VS200 acquisition software, equipped with 2x ACH and 20X XAPO objectives coupled to a 21 mm diagonal ORCA fusion sCMOS camera, C14440. The microscope was equipped with a X-Cite XYLIS, XT720L 385 nm laser, utilizing a DAPI, FITC, and mCherry LED cubes. We analyzed the images using ImageJ software to assess the expression intensity of both NET and TH following norepinephrine system lesioning. The effects of DSP-4 treatment on norepinephrine terminals were quantified by measuring the fluorescence intensity of NET and TH expression using ImageJ. We traced the outline of each individual brain section using a polygonal selection tool, allowing us to measure the mean fluorescence intensity for DAPI, NET, and TH. To account for potential variations in tissue processing and imaging conditions, we normalized TH and NET fluorescence intensity to the mean DAPI fluorescence intensity within each section. Finally, we compared the normalized values across different treatment conditions and brain regions to assess the impact of DSP-4 on norepinephrine terminal integrity.

Statistical analysis

We used factorial ANOVAs and t-tests using SPSS (IBM, version 25, GLM procedure). When we obtained significant main effects and interaction effects (p < 0.05, two-tailed), we followed them with post-hoc tests (Fisher PLSD). Because our multifactorial ANOVAs yielded multiple main and interaction effects, we only report significant effects that are critical for data interpretation. We indicate results of post-hoc analyses in the figures but do not describe them in the Results section. We indicate p values for values that are less than 0.001 as p < 0.001 and provide exact p values for values smaller than 0.05 and greater than 0.001. In Table S1 we provide a complete report of the statistical results for the data described below. In Table S2 we provide details of rats’ distribution across high, medium, and low social motivation groups. Sample sizes are similar to those reported in previous publications [27, 39–41, 55].

Results

Rats are highly motivated to engage in volitional mutual social interactions

Both humans and laboratory animals engage in social interactions by interpreting and adjusting their behaviors in response to various complex cues to ensure appropriate engagement [56, 57]. To model this crucial aspect of social interactions, we developed a volitional mutual social interaction task in Experiment 1 (Exp. 1) that actively requires both resident and partner rats to be motivated to engage in social interaction (Fig. 1A). We combined classical operant conditioning procedures within a custom-made mutual social self-administration chamber (Fig. S1). This setup is based on our previous apparatus designs [27, 30, 38–43]. For a comprehensive materials and methods details, distribution of rats across groups and a report of the statistical analyses conducted for all experiments, refer to Table S1 and S2 in the Supplementary Online Materials.

Fig. 1. Mutual social interaction model with alternate roles for social pairs.

A Model: Graphical representation of behavioral procedure showing the custom-made mutual social-self-administration chamber equipped with two social active levers (resident and partner), two discriminative cues (resident-houselight and partner-cuelight), a tone for social interaction, and a sliding door to interact socially. Resident and partner rats switched roles (indicated by the arrows). B Mutual social self-administration. Number of mutual social rewards over 6 sessions (1 h) and representative session showing lever presses responding for resident and partner rats when the resident started the task (top) or when the partner started the task (bottom). C Latency to social interaction. Average latency (min) to achieve social interaction. D Frequency of social interaction. Average social rewards earned per minute. E Social score. Subject assignment. Individual z-score measure for total mutual social rewards, average latency, and average frequency; percentages of rats classified as Low, Medium, and High Social groups (total n = 12 pairs – resident initiating the task; and 12 pairs – partners initiating the task). Data are mean ± SEM. See also Fig. S1 and S2.

Across six 1-h sessions, rat pairs increased their number of mutual social interaction rewards (Fig. 1B Left – Session F_{(5, 100)} = 29.0, p < 0.001). Social pairs quickly mutually engaged in lever responding (Fig. 1B, Right), as evidenced by the decreased latency to social interaction (Fig. 1C – Session F_{(5, 100)} = 6.6, p < 0.001) and increased frequency of social interaction (Fig. 1D – Session F_{(5, 100)} = 27.8, p < 0.001 – the maximum reward frequency is one reward every two minutes, represented by the gray dotted line). This effect remained consistent regardless of whether the pairs initiated the task as residents or partners, as well as regardless of their sexes (p > 0.05). While rodents consistently engaged in mutual social interactions, we sought to determine whether resident and partner pairs would exhibit high motivation for mutual social interactions. Thus, we calculated a social score to observe the distribution of the social pairs (Fig. 1E) based on three measures: the total number of social rewards (under Fixed-Ratio 1 (FR1) schedule of reinforcement – Fig. S2A), the average latency (Fig. S2B), and the average frequency (Fig. S2C) across the sessions. Subsequently, we computed the mean z-score for each pair and divided the sample (n = 24 pairs of rats – 12 female pairs; 12 male pairs; 12 pairs with residents initiating the task and 12 pairs with partners initiating the task) into three sub-groups: rats with mean z-scores ≥0.5 were classified as ‘high social’ pairs ( ~ 33% for both resident [n = 4; 2 F/2 M] and partner [n = 4; 2 F/2 M] pairs), rats with mean z-scores ranging from 0.4 to -0.4 were classified as ‘medium social’ pairs ( ~ 50% for both resident [n = 6; 3 F/3 M] and partner [n = 6; 3 F/3 M] pairs), and rats with mean z-scores ≤-0.5 were classified as ‘low social’ pairs ( ~ 17% for both resident [n = 2; 1 F/1 M] and partner [n = 2; 1 F/1 M] pairs).

Exp. 1 demonstrates that rats trained in an operant social task voluntarily and mutually engaged in social interactions, with varying levels of social motivation across animals within our experimental cohort. Importantly, this motivation remained consistent regardless of the rats’ roles (resident or partner) or sex. Additionally, we showed that a significant proportion of paired male and female rats exhibited high motivation to participate in voluntary mutual social interaction.

Rats remain highly motivated for social interaction at increasing effort requirements

Exp. 2 (Fig. 2A) provided an evaluation of how increased effort requirements affect the motivation of both resident and partner rats to engage in mutual social interaction. We trained male and female paired rats for mutual social interaction (Fig. S3A–C) across nine 1-h sessions, replicating the volitional mutual social interaction observed in Exp. 1. We then tested the rats’ motivation to engage in mutual social interactions at increased FR requirements for both residents and partners. The rats performed reliable mutual social interaction at different FR requirements (Fig. 2B, Left – Session X Effort: F_{(4, 40)} = 3.2, p = 0.02) by improving their mutual lever presses in response to the varying FR requirements (Fig. 2B, Middle and Right – Effort: F_{(2, 40)} = 78.5, p < 0.001). The rats increased the latency to social interaction (Fig. 2C – Effort: F_{(2, 20)} = 40.0, p < 0.001) and decreased the frequency of social interaction (Fig. 2D – Effort: F_{(2, 20)} = 29.8, p < 0.001) due to the additional time required to complete the task at higher FR. However, there was no interaction between Session and Effort or Sex (p > 0.05) even when required to lever press double or quadruple the amount to interact socially.

Fig. 2. Mutual social interaction model at increasing effort requirement.

A Model: Graphical representation of behavioral procedure showing the increased effort requirements (indicated by the arrows) for both resident and partner rats. B Mutual social self-administration. Left. Number of mutual social rewards over 9 sessions (1 h) at increasing FR (sessions 1–3 FR1; sessions 4-6 FR2, sessions 7–9 FR4). Center. Number of lever presses for both resident and partner rats over 9 sessions (1 h) at increasing FR. Right. Representative session showing lever presses responding for resident and partner rats at increasing FR requirements (Top: FR1; Middle: FR2; Bottom: FR4). C Latency to social interaction. Average latency (min) to achieve social interaction at different FR requirements. D Frequency of social interaction. Average social rewards earned per minute at different FR requirements. E Social score. Subject assignment. Individual z-score measures for total mutual social rewards, total lever presses at different FR requirements, average latency, and average frequency; percentages of rats classified as Low, Medium, and High Social groups (total n = 12 pairs). Data are mean ± SEM. See also Fig. S1, S3, and S4.

To determine the social score of each rat (Fig. 2E), we used four measures: the total number of social rewards (across FR schedules of reinforcement – Fig. S4A), the total number of lever presses for both residents and partners (Fig. S4B), the average latency (Fig. S4C), and the average frequency (Fig. S4D) across the sessions. Using the thresholds established in Exp. 1, we calculated the mean z-score for each pair and divided the sample (n = 12 pairs of rats – 6 female pairs; 6 male pairs) into ‘high social’ pairs ( ~ 43% – [n = 5; 3 F/2 M]), ‘medium social’ pairs ( ~ 41% – [n = 5; 2 F/3 M]), and ‘low social’ pairs ( ~ 16% – [n = 2; 1 F/1 M]).

Exp. 2 demonstrates that levels of social interaction remained relatively steady despite increased effort, indicating that social interaction was valuable to the rats. This suggests that the motivation to interact socially was not merely a side effect of lever pressing for motor reward or other alternative explanations. This finding ultimately models the motivated nature of mutual social interactions reported in humans [5], emphasizing the critical role of both resident and partner rats’ motivation in engaging in mutual social interactions and continuing further interactions.

Social partners are highly motivated to engage in mutual social interactions

In Exp. 3 (Fig. 3A), we investigated whether social partners exhibit motivation to engage in mutual social interactions. While studies of laboratory rats, particularly wild-type Norway rats, have shown that interactions such as food provisioning or allogrooming elicits social responses [21, 57], the measurement of motivation for mutual social interaction remains largely unexplored [23–30]. We took an approach similar to Exp. 2, using increasing FR effort requirements and a progressive ratio (PR) reinforcement schedule exclusively for the partners while maintaining an FR1 schedule for the residents.

Fig. 3. Mutual social interaction model at increasing effort requirement for social partner.

A Model: Graphical representation of behavioral procedure showing the increased effort requirements (indicated by the arrows) for partner rats. B Mutual social self-administration. Left. Number of mutual social rewards over 9 sessions (1 h) at increasing FR (sessions 1–3 FR4; sessions 4-6 FR8, sessions 7–9 FR16). Center. Number of lever presses for both resident and partner rats over 9 sessions (1 h) at increasing FR. Right. Representative session showing lever presses responding for resident and partner rats at increasing FR requirements (Top: FR4; Middle: FR8; Bottom: FR16). C Latency to social interaction. Average latency (min) to achieve social interaction at different FR requirements. D Frequency of social interaction. Average social rewards earned per minute at different FR requirements. E Progressive ratio (PR): Left. Number of mutual social rewards earned during PR tests over 3 sessions. Center. Number of lever presses for both resident and partner rats during PR tests over 3 sessions. Right. Representative session showing lever presses responding for resident and partner rats during PR tests (Top: rat1; Bottom: rat2). F Latency to social interaction. Average latency (min) to achieve social interaction during PR tests. G Frequency of social interaction. Average social rewards earned per minute during PR tests. H Social score. Subject assignment. Individual z-score measure for total mutual social rewards, total lever presses at different FR requirements and PR tests, average latency, and average frequency; percentages of rats classified as Low, Medium, and High Social groups (total n = 12 pairs). Data are mean ± SEM. See also Fig. S1, S3, and S5.

The social partners were able to engage in social interaction at different FR requirements (Fig. 3B Left) and increased their mutual lever presses in response to the varying FR requirements (Fig. 3B Middle and Right – Session X Effort X Role: F_{(4, 80)} = 4.1, p = 0.004). As reported in Exp. 2, the latency to social interaction increased (Fig. 3C – Effort: F_{(2, 20)} = 32.0, p < 0.001) and the frequency of social interaction decreased (Fig. 3D – Effort: F_{(2, 20)} = 28.2, p < 0.001) due to the increased complexity of completing higher effort requirements while engaging in mutual social interaction across FR requirements. Male and female partners were able to engage in social interactions even as the effort requirements increased within a session during the PR sessions (Fig. 3E Left – Session: F_{(2, 20)} = 7.5, p = 0.004) and they increased their mutual lever presses for mutual social interactions (Fig. 3E Right – Session X Role: F_{(2, 40)} = 7.5, p = 0.002). Furthermore, they did not change the latency to social interaction over sessions (Fig. 3F – p > 0.05) and increased frequency of social interaction (Fig. 3G – Session: F_{(2, 20)} = 4.1, p = 0.03).

When computing the social score of each pair (Fig. 3H), using data from the total number of social rewards (Fig. S5A), the total number of lever presses (Fig. S5B), the average latency (Fig. S5C), and the average frequency (Fig. S5D) during both increased FR requirements and PR schedule of reinforcement, we divided the sample (n = 12 pairs of rats – 6 female pairs; 6 male pairs) into ‘high social’ pairs ( ~ 50% – [n = 6; 4 F/2 M]), ‘medium social’ pairs ( ~ 33% – [n = 4; 2 F/2 M]), and ‘low social’ pairs ( ~ 17% – [n = 2; 0 F/2 M]).

Exp. 3 demonstrates that both male and female partners are active participants in social interaction rather than passive contributors. They are motivated to engage in mutual social interactions regardless of effort requirements. Moreover, these findings confirm that the results of Exp. 2 are not an artifact of the setup. Instead, rats act with intention, displaying goal-directed behavior to achieve social interaction.

Norepinephrine mediates mutual social interaction both systemically and centrally

We hypothesized that catecholamines modulate the motivation to engage in mutual social interactions. Specifically, attenuation of norepinephrine signaling occurs either through blockade of post-synaptic alpha-1 receptors or activation of pre-synaptic alpha-2 auto-receptors [58–60]. In Exp. 4 (Fig. 4A), we initially trained male and female paired rats for mutual social interaction (Fig. S6A–D) across nine 1-h sessions. To evaluate the role of catecholamines in motivating mutual social interactions, we conducted separate test sessions for resident and partner rats (n = 18 pairs, 9 female and 9 male). During resident testing, compounds were administered solely to the resident using a within-subject design, with the effort requirement increased to FR4 (while maintaining FR1 for the partner). In partner testing, compounds were administered to the partner with the same design, and effort was also increased to FR4 (while FR1 was maintained for the resident). Three adrenergic receptor-targeting drugs were used to investigate norepinephrine’s role in social interactions: Prazosin (alpha-1 antagonist), Clonidine (alpha-2 agonist), and Rx821001 (alpha-1 and alpha-2 antagonist). These drugs were selected based on prior research indicating their effectiveness in modulating motivated behaviors in rodents [26, 47, 61, 62]. The goal was to determine how norepinephrine signaling affects motivation for social interaction, providing insights into the neurobiological mechanisms underlying social behavior. A Latin-square design was used to prevent cumulative drug effects, ensuring that all rats received all doses (including the vehicle) across different sessions. Additionally, rats were retrained to their baseline behavior between drug testing sessions to ensure consistency.

Fig. 4. Mutual social interaction model during norepinephrine manipulations.

A Model: Graphical representation of behavioral procedure showing the pharmacological treatments (indicated by the syringes) for both resident and partner rats. B Mutual social self-administration. Number of mutual social rewards (1 h sessions) across 3 doses per each compound (Left. Prazosin; Center. Rx821001; Right. Clonidine) for resident and partner rats. C Latency to social interaction. Average latency (min) to achieve social interaction (1 h sessions) across 3 doses per each compound (Left. Prazosin; Center. Rx821001; Right. Clonidine). D Frequency of social interaction. Average social rewards earned per minute (1 h sessions) across 3 doses per each compound (Left. Prazosin; Center. Rx821001; Right. Clonidine). E Mutual social self-administration. Number of mutual social rewards over 9 sessions (1 h) after either vehicle or DSP-4 injections for resident and partner rats at increasing FR requirements (Left. FR1; Center. FR2; Right. FR4). F Latency to social interaction. Average latency (min) to achieve social interaction over 9 sessions (1 h) after either vehicle or DSP-4 injections at increasing FR requirements. G Frequency of social interaction. Average social rewards earned per minute over 9 sessions (1 h) after either vehicle or DSP-4 injections at increasing FR requirements. H Representative images and quantification. Left. Representative photomicrographs of the effect of either vehicle or DSP-4 on TH (tyrosine hydroxylase) or NET (norepinephrine transporter) on cell bodies and norephineprine terminals throughout the brain at different bregma levels (show in mm). (Scale bar: 100 µm). Right. Quantification of NET/DAPI (left) and TH/DAPI (right) expression intensity across multiple Bregma levels (mm). (I) Social score. Subject assignment. Individual z-score measure for total mutual social rewards, average latency, and average frequency after either vehicle or DSP-4 injections; percentages of rats classified as Low, Medium, and High Social groups (total n = 8 pairs vehicle; n = 7 pairs DSP-4). Data are mean ± SEM. See also Fig. S1, S6, S7, S8, and S9.

We reported a dose-dependent effect on the number of mutual social interactions with both Prazosin (Fig. 4B – Dose: F_{(2, 64)} = 21.5, p < 0.001) and Clonidine (Fig. 4B – Dose: F_{(2, 56)} = 99.2, p < 0.001), but not with Rx821001 (Fig. 4B – p > 0.05). These effects were consistent across roles (resident or partner) and sexes (p > 0.05). Additionally, these manipulations selectively decreased the number of lever presses on the tested rat (either resident or partner – Fig. S7A-B). Furthermore, Clonidine manipulation significantly increased the latency to social interaction (Fig. 4C – Dose: F_{(2, 56)} = 26.2, p < 0.001), and both Prazosin and Clonidine decreased the frequency of social interactions (Fig. 4D – Prazosin, Dose: F_{(2, 64)} = 13.3, p < 0.001; Clonidine, Dose: F_{(2, 56)} = 65.6, p < 0.001).

Based on previous literature highlighting the role of dopamine receptors in social-related behaviors [26, 58], we also investigated dopamine’s involvement in mutual social interactions by administering the D1 receptor antagonist SCH39166 and the D2 receptor antagonist Raclopride. Our findings revealed similar patterns of results when testing the effects of SCH39166 or Raclopride on mutual social interactions for both residents and partners, with a more pronounced effect on males than females (Fig. S8A – SCH39166, Dose X Sex: F_{(2, 64)} = 5.1, p = 0.009; Raclopride, Dose X Sex: F_(2,64) = 3.9, p = 0.03). This effect was also aligned with the dose-dependent effect on lever presses based on the pairs’ role (Fig. S8B – SCH39166 – R(FR4)→P(FR1), Role: F_{(1, 32)} = 40.4, p < 0.001; R(FR1)→P(FR4), Role: F_{(1, 32)} = 73.3, p < 0.001; Raclopride – R(FR4)→P(FR1), Role: F_{(1, 32)} = 54.5, p < 0.001; R(FR1)→P(FR4), Role: F_{(1, 32)} = 83.1, p < 0.001). Furthermore, rats exhibited an increase in latency to mutual social interaction when treated with Raclopride (Fig. S8C – Raclopride, Dose: F_(2,64) = 3.3, p = 0.04) and a decrease in the frequency of social reward earned over the session when treated with SCH39166 (Fig. S8D – SCH39166, Dose: F_{(2, 64)} = 6.1, p = 0.004). However, dopamine receptor manipulation also impaired food self-administration (Fig. S8E), reinforcing dopamine’s general role in motivated behaviors rather than a specific effect on mutual social interactions [63]. This finding led us to focus selectively on the role of central norepinephrine in the next experiment.

While investigating potential pharmacological mechanisms, we also aimed to delve into other neurobiological systems beyond catecholamines, specifically targeting the serotoninergic and glutamatergic systems due to their known ability to modulate social behaviors [64, 65]. We observed no changes in mutual social interactions when rats were treated with either Ketanserin (5-HT2 antagonist) or LY341495 (mGluR2 metabotropic antagonist) (Fig. S8F-I – p > 0.05).

Building on the effects observed from systemic selective manipulation of norepinephrine, we aimed to explore the effects of central norepinephrine ablation on the motivation for mutual social interactions and the distribution of social pairs based on their social score. For this purpose, we employed N-(2-chloroethyl)-N-ethyl-2-bromobenzylamine hydrochloride (DSP-4). DSP-4 selectively targets locus coeruleus norepinephrine terminal neurons without affecting other norepinephrine neurons in the lateral tegmental nuclei [53]. Two weeks after a single i.p. injection of either vehicle or DSP-4, we assessed mutual social interaction by gradually increasing the fixed ratio (FR) effort requirements for both resident and partner rats over a two-week period. Testing began two weeks after injection, with FR demands increasing every three days, allowing one day of rest between each increase. Two weeks was selected as this is the reported time frame for DSP-4 to exert its neurotoxic effects [53]. We found that while the social pairs in the vehicle group maintained a reliable mutual social interaction, those in the DSP-4 group decreased the number of mutual social rewards over sessions in response to the effort requirements (Fig. 4E – Effort X Session X Group: F_{(4, 44)} = 2.7, p = 0.04). This effect was further supported by an increase in the latency to social interaction (Fig. 4F – Effort X Session X Group: F_{(4, 44)} = 4.2, p = 0.006) and a decrease in the frequency of social interaction (Fig. 4G – Effort X Session X Group: F_{(4, 44)} = 2.9, p = 0.03) for the DSP-4 treated social pairs. Using immunohistochemistry, we confirmed that the manipulation almost completely removed the norepinephrine transporter (NET) in norepinephrine terminals throughout the brain, while leaving intact the norepinephrine cell bodies identified using tyrosine hydroxylase (TH) (Fig. 4H). Moreover, DSP-4 selectively lesioned locus coeruleus projections (cortical regions, hippocampus, basolateral amygdala) but not lateral tegmental nuclei projections (bed nucleus of stria terminalis, central amygdala, hypothalamus) (Fig. S9). Using approaches similar to that of Exp. 1–2, we calculated the social score for each social pair, comparing vehicle (n = 8 pair of rats – 4 female pairs; 4 male pairs) and DSP-4 (n = 7 pair of rats – 3 female pairs, 4 male pairs) treated pairs (Fig. 4I). We derived the score from data on the total number of social rewards (Fig. S10A), average latency (Fig. S10B), and average frequency (Fig. S10C). We then categorized the sample into ‘high social’ pairs (Vehicle: ~38% – [n = 3; 2 F/1 M]; DSP-4: ~14% – [n = 1; 1 F/0 M]), ‘medium social’ pairs (Vehicle: ~50% – [n = 4; 1 F/3 M]; DSP-4: ~43% – [n = 3; 2 F/1 M]), and ‘low social’ pairs (Vehicle: ~12% – [n = 1; 1 F/0 M]; DSP-4: ~43% – [n = 3; 0 F/3 M]).

Exp. 4 shows that disruption of the catecholamine system influences mutual social interactions. Specifically, targeted central ablation of the norepinephrine system leads to a significant reduction in the motivation to engage in mutual social interactions. This perturbation results in a decrease of over half in the proportion of highly motivated pairs compared to vehicle controls.

Central norepinephrine manipulation via DSP-4 does not affect palatable food self-administration

In Exp. 5 (Fig. 5A), our goal was to evaluate whether the effects observed in Exp. 4 were specific to mutual social interactions or if they generalized to a different natural reward (palatable food pellets). We aimed to investigate the possibility of general impairments rather than a selective effect on mutual social interaction, considering the potential for significant changes in the rats’ level of arousal or attention following norepinephrine manipulations [66]. We trained non-food-restricted male and female rats for palatable food self-administration over nine 2-h sessions with increasing fixed-ratio (FR) effort requirements, to use similar training and testing protocols used for volitional mutual social interaction (n = 12 pairs of rats – 6 female pairs; 6 male pairs). Following stable palatable food self-administration (Fig. S11A), we tested the effect of systemic administration of either Prazosin or Clonidine on food self-administration. We found no significant effect of either dose of Prazosin on palatable food self-administration, as evidenced by the number of pellet rewards (Fig. 5B, Left – p > 0.05) and lever presses, with the rats distinguishing between active and inactive levers (Fig. 5B, Right – Lever: F_{(1, 16)}: 35.0, p < 0.001). However, when we tested the rats with Clonidine, we observed a decrease in the number of pellet rewards (Fig. 5C, Left – Dose: F_{(2, 14)}: 7.4, p = 0.006) and lever presses (Fig. 5C, Right – Dose: F_{(2, 14)} = 13.1, p < 0.001). Notably, this effect was driven by decreasing palatable food self-administration at the high dose (p = 0.002) but not at the low dose (p > 0.05). Next, we tested the effect of lesioning norepinephrine-containing terminals in the brain using DSP-4 injection on food self-administration using the same procedure described in Exp. 4. We did not detect any changes in palatable food self-administration at any of the FR effort requirements (Fig. 5D, Left – p > 0.05), with rats increasing the number of lever presses in response to the FR requirements (Fig. 5D, Middle and Right – Effort X Lever: F_{(2, 16)} = 25.2, p < 0.001). Immunohistochemistry confirmed that DSP-4 manipulation ablated the norepinephrine transporter (NET) throughout the brain (Fig. 5E). Moreover, as reported in Exp. 4, DSP-4 selectively lesioned locus coeruleus projections (Fig. S12).

Fig. 5. Alternative natural reward (food) motivation during norepinephrine manipulations.

A Model: Graphical representation of behavioral procedure palatable food self-administration chamber equipped with one active lever (food) one inactive lever, one discriminative cues (red light), one conditioned stimulus (yellow light), and a food receptacle to get palatable pellets. B Food self-administration. Left. Number of food rewards (2 h sessions) across 3 doses of Prazosin. Right. Lever presses. Number of active and inactive lever presses (2 h sessions) across 3 doses of Prazosin. C Food self-administration. Left. Number of food rewards (2 h sessions) across 3 doses of Clonidine. Right. Lever presses. Number of active and inactive lever presses (2 h sessions) across 3 doses of Clonidine. D Food self-administration. Left. Number of food rewards over 9 sessions (1 h) after either vehicle or DSP-4 injections at increasing FR requirements (Left. FR1; Center. FR2; Right. FR4). Center and Right. Number of active and inactive lever presses over 9 sessions (1 h) after either vehicle (center) or DSP-4 (right) injections at increasing FR requirements. E Representative images and quantification. Left. Representative photomicrographs of the effect of either vehicle or DSP-4 on TH (tyrosine hydroxylase) or NET (norepinephrine transporter) on cell bodies and norephineprine terminals throughout the brain at different bregma levels (show in mm). (Scale bar: 100 µm). Right. Quantification of NET/DAPI (left) and TH/DAPI (right) expression intensity across multiple Bregma levels (mm). Data are mean ± SEM. See also Fig. S1, S10 and S12.

Exp. 5 demonstrates that ablating norepinephrine-containing terminals in the brain does not impair motivation for a natural reward, such as palatable food pellets. Combined with findings from Exp. 4, this suggests that DSP-4 treatment selectively affects mutual social interactions rather than general reward processing. Furthermore, our data indicate that this selective effect is mediated by the α1 norepinephrine receptor. Supporting this conclusion, converging evidence shows that both α1 receptor antagonism and low-dose α2 receptor agonism specifically disrupt mutual social interactions while preserving food reward motivation in both male and female rats. These findings highlight norepinephrine signaling as a key regulator of social motivation, independent of the drive for non-social natural rewards.

Discussion

We developed a volitional mutual social interaction model where same-sex pairs of rats actively engage in affiliative social behavior. Our findings show that rats are highly motivated to interact socially, with low latency and high frequency of interactions, regardless of their role, sex, or training conditions. Analyzing behavioral data through average social score, we found that rats remained motivated for mutual social interactions, even when effort requirements were increased for one or both rats. Additionally, we demonstrated the involvement of the norepinephrine system using systemic pharmacological manipulations and central lesioning. Disrupting norepinephrine specifically reduced the motivation for mutual social interactions without affecting motivation for food rewards.

The significance of relevant animal models in social neuroscience

The field of social neuroscience is advancing toward animal models that reflect complex, evolutionarily conserved behaviors with significant translational value [67, 68]. Rodents, as inherently social species, engage in nuanced interactions that require interpreting social cues such as facial expressions, whisker movements, odors, and vocalizations [55, 69, 70]. Social interactions are rewarding in both humans and rodents [71], yet existing models are limited, partly due to assumptions that social cooperation is rare and primarily driven by self-interest [5, 7]. However, growing evidence challenges this view [72, 73], showing that rodents engage in prosocial behaviors without immediate personal gain, suggesting the presence of generalized empathy or altruistic-like behaviors [15, 18, 22, 74, 75]. Our model uniquely measures the motivation of two individuals to seek social rewards, offering critical insights into social interaction mechanisms [76]. Unlike traditional models that limit the partner’s role to passive interaction, our custom-designed apparatus differentiates between social-seeking and exploratory behaviors [23–30]. While our approach relies on operational measures such as lever presses and interaction latency, capturing the full complexity of rodent social behavior requires integrating computational neuroethology tools such as SimBA [77], DeepLabCut [78], and SLEAP [79]. These tools enable predictive classifiers for nuanced behavioral analysis, including ultrasonic vocalizations (USVs), which play a critical role in rodent social communication [80–85].

Our study provides new insights into rodent social behavior, showing that rats are highly motivated to seek social interactions. The proportion of highly motivated social pairs increased as task demands rose, from about 33% to 43% when both rats faced higher effort requirements. This proportion further rose to approximately 50% when social partners had to exert greater effort for mutual interaction under a progressive ratio schedule, even when the resident rat’s task remained low-effort. These findings suggest a strong mutual drive for social interaction, consistent with previous studies showing that rodents continue seeking social interaction after training, even under extinction conditions when physical interactions are unavailable [30, 38, 55]. To maintain high motivation for social interaction, we single-housed rats throughout the experiments [27, 38]. Our findings show that social self-administration is higher in single-housed rats than in paired-housed rats and that motivation is stronger for familiar versus unfamiliar partners in single-housed rats but not in paired-housed rats [86]. Based on these findings, we predict that motivation for mutual social interaction in our model will depend on partner familiarity and housing conditions. This setup provides a dynamic platform to manipulate these factors and assess individual differences. Additionally, while we have observed strain differences in social motivation among mice [30, 43, 87], we have not found similar differences across rat strains [38]. Therefore, we predict that mutual social interaction patterns will remain consistent across different rat strains. Finally, our findings align with studies on reciprocal cooperation, which show that the willingness to assist cooperative partners increases significantly with rising behavioral costs, compared to assistance toward less cooperative rats [21, 72].

Neurobiology mediating mutual social interactions

To explore a possible mechanistic explanation on the high motivation for engaging in mutual social interactions, we examined the contribution of the catecholamines dopamine and norepinephrine [88]. While different species may display varying cognitive mechanisms when demonstrating altruism-like behaviors [15], the complexity of social behavior is regulated by an evolutionarily conserved brain network [31, 32]. Catecholamines have been implicated in various aspects of social behaviors [58], spanning from social interaction to social play behaviors, due to their well-known role in incentive motivation, attention, and arousal processes [63, 66].

Our data indicate that systemic pharmacological manipulation of norepinephrine—either through α1 receptor antagonism or α2 receptor agonism—selectively impairs the motivation of social pairs to engage in mutual social interactions, regardless of their role in the task. Notably, these effects were specific to social behavior and did not impact food self-administration. The selective disruption of mutual social interaction, along with a significant reduction in the proportion of highly motivated social pairs from approximately 38% to 14%, suggests a critical role for brain norepinephrine. This effect likely originates from one or a combination of the widespread projections of the locus coeruleus, which is the primary source of norepinephrine in the brain [66]. Locus coeruleus-derived norepinephrine is theorized to generate a global prediction error signal, facilitating the appropriate processing of sensory information [89]. Our findings align with previous research demonstrating norepinephrine’s essential role in regulating social behaviors in rodents [58, 90], further supporting its involvement in motivation for social interaction. Social interactions are complex processes that require interpreting signals and adjusting behaviors, with the norepinephrine system playing a key role in humans [34]. From a translational perspective, our data suggest that a noradrenergic mechanism acts as an internal regulator during social behavior. Disruption of this system impairs rats’ ability to seek or provide appropriate social responses to their partners.

Concluding remarks

Using operant and behavioral learning approaches, we found that highly social pairs increase as effort demands rise. This highlights the need to study social behaviors in inbred rat strains, ensuring key participants are included, especially in naturalistic settings [72, 73]. Translationally, our model aids in understanding motivation in social interactions, offering a platform to explore neurobiological mechanisms and their relevance to human behavior [91].

Author contributions

CAL and MV conceptualized the project and designed the experiments; CAL, AC, BF, JG, RM, KMP, ZH, APS, LAR, and MV collected the data; CAL and MV analyzed the data; CAL, AC, BF, JG, RM, KMP, ZH, APS, LAR, and MV contributed to different aspects of the write-up of the paper.

Funding

The research was supported by funding from intramural research program of NIDA (RM, LAR), a grant from NIDA [DA047976] (MV) and NIMH [MH129310] (MV).

Data availability

Materials, datasets, and protocols are available upon request to Marco Venniro (marco.venniro@som.umaryland.edu).

Competing interests

The authors declare no competing interests.

Supplementary information

The online version contains supplementary material available at 10.1038/s41386-025-02113-3.