Social relationships are critical for health and wellbeing of the individual and for survival of social species. Thus, social behaviors are valued and motivating, while social isolation and ostracization are aversive and stressful. While different species exhibit a variety of behaviors that facilitate social bonding and maintain social relationships, many of the neural mechanisms that govern the display of these behaviors are conserved across species. For example, neuropeptides such as oxytocin (OT) and vasopressin, and neurotransmitters such as dopamine (DA) and endorphins, are critically involved in promoting affiliative and social-seeking behaviors. In addition, stress-related neuropeptides such as glucocorticoids and corticotrophin releasing hormone (CRH) are increased during social isolation and social ostracization, leading to anxiety- and depressive-like behaviors and altering sociality in various ways. This review will synthesize what we currently know about the neurobiological foundations of loneliness in humans and the use of animal models of social isolation to provide a more mechanistic approach to study how the brain adapts to being alone. Additionally, we will discuss social loss as a unique form of loneliness and isolation in both humans and animal models and how the neurochemical underpinnings of loss overlap with and diverge from those that influence loneliness in the absence of loss.

From an ecological standpoint, species that have evolved to rely on sociality and community for survival should certainly have neurobiological mechanisms in place that drive individual members of the species to seek out and form relationships with others. In such species, social groups provide greater protection from predators (i.e., shared vigilance, mob defense), enhances success in locating and maintaining access to resources including food and mates, and increases survival of offspring by increasing the number of caregivers and protectors (Lee, 1994; Burkart and van Schaik, 2016; van Schaik, 2016). Socialization in these species begins early in development, and the social environment (and thus, the definition of what it means to be appropriately social at that time point) change throughout the lifespan (Box 1). Social interaction is driven neurobiologically by activation of reward circuitry involving the neurotransmitter DA being synthesized from the midbrain ventral tegmental area (VTA) and released into the ventral striatum (i.e., nucleus accumbens, NAc) and higher-order cortical regions (i.e., prefrontal cortex, PFC; insular cortex, IC; anterior cingulate cortex, ACC). Through this mechanism and via coordination with the neuropeptides OT and vasopressin that are released during social contact and intimacy, forming and maintaining social relationships is gratifying and valuable. Many social mammals show affiliation and affection through the physical touch of social grooming (Carter et al., 1999). In non-human mammals, social grooming causes a release of endogenous opioids (i.e., endorphins; Keverne et al., 1989), neurochemicals that promote reward and reduce pain sensitivity. This form of tactile stimulation is encoded by mechanoreceptive C-tactile (CT) fibers at the base of most hair follicles in many species (rodents, Leem et al., 1993; primates, Kumazawa and Perl, 1977; humans, Vallbo et al., 1999; Löken et al., 2009; Olausson et al., 2010). These axons are unmyelinated, respond specifically to slow and gentle stroking (∼2.5 cm per second), and directly trigger endorphin release (Nummenmaa et al., 2016). As humans no longer have fur to socially groom, these receptors respond instead to stroking, caressing, touching, and hugging which we use as a means of expressing and strengthening intimate relationships (Dunbar, 2010; Löken and Olausson, 2010; Suvilehto et al., 2015). Even non-tactile behaviors such as laughing (Caruana, 2017; Manninen et al., 2017), singing (Pearce et al., 2015), dancing (Tarr et al., 2015, 2016), and emotional story telling (Dunbar et al., 2016) in social contexts can facilitate the release of endogenous opioids. Thus, showing and receiving affection and intimacy activates neurobiological processes that reward and promote continued display of such behaviors.

Being close to other individuals can also be stress buffering and protective against the consequences of hardship and adversity. Emotionally, social support is related to more positive moods (Williamson et al., 2019), better mood regulation (Tabatabaei et al., 2018), and lower psychological distress (Lepore et al., 1991; Lee C. Y. S. et al., 2018). Positive social experiences shape our self-perceptions, and thus have a great impact on how we view ourselves, our merit, and our ability to overcome pain or failure. Central to this idea is self-compassion, which has three main components: (1) self-kindness; (2) common humanity, or the view of oneself as part of a broader human experience; and (3) mindfulness of one’s own thoughts and feelings (Neff, 2003). Because our own self-esteem and self-view depends on our perceived feelings of social value and inclusion (Leary and Baumeister, 2000), social support and positive social interactions encourage a more balanced, self-forgiving perspective that may facilitate better emotional well-being. In support, people with high perceived social support have more positive and fewer negative self-perceptions and beliefs about others’ views of themselves (Sarason et al., 1991). Social support is also positively associated with mindfulness, self-compassion, and savoring (the use of thoughts and actions to increase appreciation of positive experiences and emotions), and these three factors significantly mediate the relationship between perceived social support and increased psychological well-being (Wilson et al., 2020). By promoting a more positive view of our ability to navigate hardship, social support may also affect stress-responsivity when experiencing distress. Indeed, social support suppresses cortisol release in humans given a social stress test or experiencing stressful events (Heinrichs et al., 2003; Raymond and Sheppard, 2017), predicts lower cardiovascular reactivity to stress (Howard et al., 2017), and increases pain tolerance (Roberts et al., 2015). Studies in non-human primates also support the stress-buffering role of social affiliation. Introducing a number of aversive stimuli that would typically increase stress responsivity does not do so when animals are presented the stimuli in social groups (Coe et al., 1982; Stanton et al., 1985; Winslow et al., 2003). Stress-buffering can even occur in the absence of physical contact, as exposure to vocalizations of a pair-mate was sufficient to blunt the increase in urinary cortisol of isolated male marmosets (Rukstalis and French, 2005). Studies in rodents have also shown a protective effect of social connectedness against the development of adverse behavioral and physiological consequences. For instance, socially housed rats born to a lineage of stressed mothers (multigenerational ancestral stress) exhibited blunted HPA axis response to chronic stress compared to those that were isolated from peers during stress (Faraji et al., 2017). In the socially monogamous prairie vole (Microtus ochrogaster), stress responsiveness is similarly buffered when an animal experiences a stressor in the presence of his or her pair-bonded partner (Smith and Wang, 2014; McNeal et al., 2017; Donovan et al., 2018). Taken together, research from human and animal studies highlight the cognitive, behavioral, and physiological benefits of social affiliation, which implies that removal of such support could have negative repercussions for individual health and well-being.

While social affiliation is rewarding and valued, social isolation and ostracization are stressful and aversive. Thus, individuals that perceive being socially isolated, regardless of actual social connectedness, experience feelings of loneliness that are severely distressing. Because loneliness is characterized by the perceived state of isolation, it is unsurprising that lonely individuals have more negative perceptions of everyday interpersonal interactions (Hawkley et al., 2007). This tendency to view social interactions through a negative lens may be driven by previous experience with social relationships. For example, poor-quality social relationships (Mullins and Dugan, 1990; Hawkley et al., 2008) and conflict in current marital or family relationships (Jones D. C., 1992; Burke et al., 2012) are risk factors that predict feelings of loneliness and isolation (Hawkley et al., 2008). As a result, loneliness is also related to stronger expectations of, and motivation to avoid, negative social outcomes and weaker expectations of positive social outcomes (Gable, 2006) as well as increased comorbidity with symptoms of social anxiety (Lieberz et al., 2021a). Because of this fear of social rejection, loneliness is also associated with increased hypervigilance to social threats and, therefore, lonely individuals are more likely to be less trusting and more hostile in uncomfortable social situations (Duck et al., 1994; Anderson and Martin, 1995; Cacioppo and Patrick, 2008; Lieberz et al., 2021b). Despite the tendency to anticipate social rejection, lonely individuals still desire social interaction, behave in ways to promote social connectedness (DeWall and Richman, 2011), and show increased sensitivity to social information (Pickett et al., 2004; Gardner et al., 2005). However, this heightened sensitivity during social interactions, driven by the need to feel socially connected and biased by experience with “failed” relationships, may ultimately prime the expectation of negative outcomes once attempts at social connection are initiated, leading to reduced fulfillment and positive affect following positive social interactions (Lieberz et al., 2021b). Furthermore, this series of events produces anxiety surrounding the social environment, resulting in physiological and health-related consequences that will be summarized in the following section.

The experience of loneliness is extremely distressing, leading to changes in stress physiology and reactivity in lonely individuals. Stress reactivity is governed by the body’s hypothalamus-pituitary-adrenal (HPA) axis, where stress-related neuropeptides are released from the hypothalamus, causing excretion of “releasing factors” from the pituitary to signal glucocorticoid (cortisol in humans and non-human primates, corticosterone in rodents) release from the adrenal glands. This process is regulated by a negative feedback loop – cortisol released by the adrenal gland will elicit a behavioral response (in concert with the autonomic fight or flight) toward environmental stimuli, but it will also bind to receptors in the brain that turn off the HPA axis which dampens continued cortisol release. Disruptions in both the basal functioning of the HPA axis as well as its ability to appropriately activate and deactivate under stressful conditions have been shown to underlie a variety of psychopathologies (reviewed in Buitelaar, 2013), and similar disruptions can be found in lonely individuals. For instance, lonely people have higher salivary (Steptoe et al., 2004; Doane and Adam, 2010), plasma (Pressman et al., 2005), and excretory (Kiecolt-Glaser et al., 1984) cortisol levels. Trait loneliness is also associated with altered diurnal pattern of cortisol release characterized by lower levels in the morning and afternoon and much higher levels in the evening (Cacioppo et al., 2000). Interestingly, lonely students in this study were just as likely to engage in restorative behaviors (sleep, exercise, spending time with others) as non-lonely students; yet these activities did not seem to blunt stress reactivity as it did in the socially embedded individuals. Basal heart rate and heart rate reactivity after reciting both a social and a non-social speech was also lower in lonely individuals compared to non-lonely individuals, and this suggests blunted autonomic nervous system functioning (Cacioppo et al., 2000). Research in socially deprived chimpanzees has also shown increased HPA and stress reactivity, and resocialization caused a reversal of isolation/deprivation-induced increase in glucocorticoid secretion (Reimers et al., 2007).

Loneliness can increase sensitivity to social disconnection, impacting the emotional response to social isolation and the expectations about others. Thus, the perception of isolation may reflect functional changes to brain regions involved in processing social information, regulating and expressing emotion, and assigning reward value to social interactions. In support, the reward-related ventral striatum showed reduced activation in response to viewing images of unfamiliar people in lonely compared to non-lonely individuals (Cacioppo et al., 2009). Furthermore, Inagaki et al. (2016a) found that lonely people had higher activity in the ventral striatum when viewing pictures of familiar people compared to strangers. Thus, it is possible that, due to increased negative expectations and higher anxiety of social outcomes, lonely individuals have a bias against novel social encounters and toward interactions with people they know will respond positively to their attempts for social connection. Moreover, the insula and anterior cingulate cortex are typically activated during presentation of social-related words (i.e. fight, glory) compared to individual words (i.e. talent, pain; Arioli et al., 2021). However, activity in both regions is dampened in lonely individuals during resting state (Morr et al., 2021) while activity in just the insula is blunted in response to viewing pleasant non-social stimuli (Cacioppo et al., 2009) and during participation in a social trust task (Lieberz et al., 2021b). Gray and white matter volume is also lower in the insula and PFC of individuals scoring high on loneliness, which could indicate reduced myelination and/or synapse number that would reduce signal efficacy (Nakagawa et al., 2015; Düzel et al., 2019). While activity is dampened in lonely individuals at rest and during a variety of positive social and non-social stimuli, the insula, ACC, PFC, and ventral striatum are more active in non-lonely individuals that reported feeling excluded during a virtual game (Eisenberger et al., 2003, 2007), and this activity pattern in response to exclusion predicts loneliness scores (Oh, 2020). Thus, it could be that these regions are sensitive to particular information in lonely individuals, i.e., less activation during pleasant stimuli but more activation during aversive or unpleasant stimuli. Furthermore, activation of visual attention regions following presentation of negative social words was greater for lonely compared to non-lonely individuals (Cacioppo S. et al., 2015), suggesting increased sensitivity to negative social stimuli. Altered emotion perception may also be a risk factor for feeling lonely and isolated, as alexithymia (a personality trait characterized by impaired emotional awareness and interpersonal relating) modulates the correlation between insular activity and loneliness (Morr et al., 2021). The insular cortex also integrates sensory input and links these cues with emotional state of both the self and others (Davidson and Irwin, 1999; Ebisch et al., 2011; Zaki et al., 2012; Gu et al., 2013) and is a critical hub for coordinating network switching (Sridharan et al., 2008). Because attention to social cues is heightened in lonely individuals, loneliness may be characterized by disrupted ability to process and relate such social cues to emotional state of social conspecifics, leading to a feeling of unfulfillment after attempts to socially connect (Box 2).

As an acute response, social isolation should drive an individual to seek out social relationships analogous to the way hunger drives an individual to seek out and consume food. However, the findings synthesized above describe a dampening of motivation-related brain regions. Importantly, recent work has begun exploring the effects of acute social isolation in humans to better understand the shift in brain function and resulting behavioral output from acute to chronic isolation and loneliness. Interestingly, Tomova et al. (2020) found that acute social isolation induces midbrain activation that they likened to craving, as similar responses occurred in people that had fasted. Additionally, social and food craving led to activation of different subregions of the striatum, with social cues evoking responses in the caudate and food cues evoking responses in the nucleus accumbens (Tomova et al., 2020). Finally, they found that the increase in VTA activity and self-reported craving in the socially isolated group was blunted in individuals that scored high on chronic loneliness, suggesting that there is indeed a functional and behavioral shift that occurs when acute social isolation becomes chronic loneliness.

With regards to stress-related brain regions, findings generally indicate increased activation of the amygdala and hippocampus. For example, there is a positive correlation between amygdala volume and social distress score, and this effect was significantly mediated by loneliness (Tian, 2016). Social exclusion also increases activation of the amygdala, and this effect is greater in lonely individuals (Oh, 2020). Increased amygdala activation may be related to attention to socially salient stimuli. For example, socially embedded individuals display stronger coupling of the limbic system with the cortical salience network (which includes the ACC and IC; Bressler and Menon, 2010), while lonely individuals show reduced functional connectivity between these networks (Layden et al., 2017). Oxytocin may be involved in strengthening this connectivity in socially enriched individuals, as OT administration enhances the functional coupling of the amygdala with the cortical salience network during positive social interactions (Rilling et al., 2018). Furthermore, social network size and complexity is positively associated with amygdala volume and functional connectivity of the amygdala to both the affiliation network and the perception network (Bickart et al., 2012). This may reflect an increase in perceptual ability, particularly to affect-laden stimuli, required to foster and maintain a high number of social contacts (for reviews, see Adolphs and Spezio, 2006; Bickart et al., 2014). Because lonely individuals have heightened attenuation to emotional stimuli (Cacioppo et al., 2009), this could explain the increased amygdala activation. The amygdala also responds to threatening stimuli, which may be how lonely individuals perceive social cues based on their hypervigilance and heightened attention during social interaction (Pickett et al., 2004; Gardner et al., 2005). Moreover, reduced connectivity of the amygdala to higher-order cortical regions of the cingulo-opercular and salience networks could indicate an inability of these regions to influence or override the hypervigilance and threat sensitivity that accompanies loneliness. Hippocampal activity is also heightened during a social exclusion task (Eisenberger et al., 2007), and this region hosts a large density of glucocorticoid receptors that act as a negative feedback loop to prevent excess cortisol release during a stressful event (van Haarst et al., 1997). Together, amygdalar and hippocampal sensitivity in lonely persons may reflect their hypervigilance and stress-reactive state.

Several reviews have been published identifying the neurobiological correlates for loneliness in humans and the behavioral and neurophysiological effects associated with social isolation in rodent models (Cacioppo J. T. et al., 2015; Mumtaz et al., 2018; Tomova et al., 2021). As mentioned previously, it is difficult to directly compare the two literatures because of the largely subjective nature of loneliness in humans. However, here we will highlight some similarities in the neural correlates and brain regions affected by both experiences and will draw from the more detailed knowledge of the neurochemical underpinnings of isolation garnered from the animal research to make hypotheses about their contribution in the human loneliness condition.

One important aspect of human loneliness that is conserved in animal models using social isolation is the high-stress state that is produced as a result of being either subjectively or objectively isolated. In humans, loneliness is associated with increased social anxiety (Lieberz et al., 2021a), heightened attention to social cues (Pickett et al., 2004; Gardner et al., 2005) and increased amygdala activity and gray matter volume (Tian, 2016). In adult prairie voles, chronic social isolation reduced the total number and the overall length of dendritic branches while increasing the spine density in interneurons within the BLA (Hylin et al., 2022). These results suggest a possible reduction in the number of neuronal signals received via reduced dendritic matter and a simultaneous increase in synaptic efficacy of the signals that are received via increased number of synapses. This could indicate a biasing or “tuning in” of particular BLA neurons to specific signals being received from certain brain regions. One projection that could be biased is that of DA neurons coming from the DR, as they are especially active during isolation. In fact, DR DA neurons almost exclusively project to limbic regions, and those that terminate in the amygdala drive social behavior and negative affective state in isolated rodents. OT may also be involved, as plasma levels are increased during chronic social isolation in rodents (Hylin et al., 2022), and in humans OT modulates attention to both positively and negatively valenced social cues (Domes et al., 2007; Gamer et al., 2010). OT neurons also project to dopaminergic neurons in both the VTA (Hung et al., 2017) and, although to a lesser extent, the DR (Otero-García et al., 2016; Fujisaki et al., 2020). Silencing PVN OT neurons in juvenile rats that have been socially isolated for 1 week restores social novelty preference, indicating that over-activation of these neurons during isolation may drive input to DA neurons to increase their activity and alter social behavior (Musardo et al., 2021).

In lonely humans, reward-related brain regions show reduced activation, which may seem counter to the above findings in rodents indicating heightened activity of DAergic regions; however, acute isolation in humans does increase VTA activity in response to social cues, although when loneliness was factored into these analyses the increase in VTA responsiveness to social cues was blunted in lonely individuals (Tomova et al., 2020). Thus, it is unclear whether DA activity remains heightened as acute social isolation becomes chronic or as objective isolation becomes subjective (i.e., loneliness). Excessive DA release in the dorsal striatum does reduce social behavior and social exploration in adult non-isolated mice (Lee Y. et al., 2018). Additionally, continuous over-activation of DAergic system over a long period of time would likely cause alterations in receptor expression or responsiveness, which could affect the strength and efficiency of the signal being received by downstream DAergic target regions. To this point, some downstream targets of DA are more active in lonely humans (amygdala and hippocampus) while others are less active (ACC, insula). If overall DA activity is heightened during loneliness and social isolation, this could simultaneously increase the desire to interact socially while promoting increased attention and hypervigilance during said social encounters, which may alter the social experience to provide less reward value. Furthermore, while human loneliness is associated with reduced activation of reward regions in the brain both at resting state and during presentation of social stimuli, activity in these regions is increased during presentation of familiar faces (Inagaki et al., 2016a). An interesting animal correlate to this behavior is one that researchers typically interpret as a social learning deficit, characterized by the inability of a rodent that has been socially isolated to spend more time with a novel animal over an animal they have already interacted with (Shahar-Gold et al., 2013; Lavenda-Grosberg et al., 2021). One could alternatively interpret this behavior as a familiarity preference rather than a lack of novelty preference. If an individual is anxious about social encounters, they may be more likely to interact with the social stimulus that they already know rather than a novel one that they have not encountered before. Alternatively, the social homeostasis model (reviewed in Lee et al., 2021) postulates that after continued attempts to engage in social behavior during acute isolation, animals will eventually switch from an active coping strategy to a passive one, leading to an adjustment in homeostatic set-point for social contact. In other words, the neural circuits that promote social behavior during acute isolation (driving an animal to get back to its social set point by restoring social connections) are the same ones that promote antisocial behavior during chronic social isolation (driving an animal to maintain it’s new social set-point by avoiding social conspecifics). Thus, reduced social novelty preference during chronic isolation may instead be the result of homeostatic mechanisms geared toward maintaining the new social set-point that is now characterized by a social deficit.

In mice, one brain region that drives novelty preference is the interpeduncular nucleus (IPN), as photoactivating this region reduces exploration of a novel social stimulus, mimicking familiarity, while reducing activation of the IPN during social conditioned place preference increases the rewarding properties of a familiar social encounter (Molas et al., 2017). As a novel social stimulus becomes familiar, activity within the IPN increases via cholinergic-glutamatergic projections from the medial habenula (mHb) to the IPN, whereas during an encounter with a novel stimulus, VTA-IPN projections activate D1 receptors to increase the salience of the novel stimulus to promote exploration. Post-weaning social isolation blunts activation of the Hb in response to a novel social encounter (Ahern et al., 2016); thus, reduced Hb activity resulting from isolation may prevent the encoding of a novel stimulus as familiar in order to bias interaction with familiar social conspecifics. The Hb and IPN have been understudied in lonely humans; however, novelty is also detected by higher order brain regions within the cognitive control networks, many of which are blunted in activity in lonely individuals (Cacioppo et al., 2009; Inagaki et al., 2015; Wong et al., 2016). These cognitive control networks, including the frontoparietal and cingulo-opercule networks, are also involved in conflict-detection, goal-oriented behavior, and behavioral flexibility, and disruptions in the ability to perform such tasks could underly some of the behavioral phenotype featured during loneliness.

Finally, the human and animal literatures could benefit from a more cohesive research trajectory. Although methodologically difficult, future research in humans could attempt to identify neurotransmitter and/or neuropeptide systems that are affected by perceived social isolation, either by measuring plasma or CSF concentrations of these neurochemicals or using PET ligands. Meanwhile, animal studies can begin further characterizing the neural circuits that regulate the behavioral phenotypes resulting from various lengths of isolation, particularly related to regions that show altered activity or connectivity in lonely humans. Additionally, a more specific time course could be identified characterizing the shift from normal, sporadic feelings of loneliness (and isolation) to pathological and chronic loneliness. Lee et al. (2021) make the point that the behavioral shift that occurs between acute and chronic isolation may be mechanistically linked to one or more variables other than time, including effort expended or number of rejected attempts. Assessing more specific factors that contribute to or influence feelings of loneliness (or the behavioral correlates in animals) is an area that would be interesting for both human and animal researchers to explore.

Losing a relationship happens naturally in social species. A relationship may dissolve due to conflict, when one individual leaves to join a new social group, or even from death of a loved one. We know that the response of an individual to social loss depends on the strength of the relationship and the type of relationship (Smigelsky et al., 2019; Sekowski and Prigerson, 2021); however, the mechanisms involved in the behavioral and physiological response to loss is poorly understood. Additionally, examining how the brain adapts to loss and overcomes the negative consequences of loss to allow an individual to form new relationships could provide valuable insights for recovery during grief and loss in humans.

This review comes at a particularly salient moment where many individuals across the globe are experiencing isolation from peers and loved ones due to the COVID-19 pandemic. As such, loneliness has been prevalent over the last two years as people have been self-isolating or quarantining due to stay at home orders (Killgore et al., 2020; Li and Wang, 2020; van Tilburg et al., 2021), and even populations that are not typically at risk for feeling lonely (i.e., students) are suffering from consequences of isolation (Bu et al., 2020). While many individuals are experiencing a deficit in the quality and quantity of social relationships during the COVID-19 pandemic, a recent meta-analysis indicates that a variety of psychological interventions are successful in reducing feelings of loneliness (Hickin et al., 2021). However, another meta-analysis revealed that interventions that simply attempted to increase social support or social contact were less effective than those that attempted to target the maladaptive cognitions that foster loneliness (Masi et al., 2011), suggesting that individuals may need to take a more active role in breaking the cycle of loneliness. Thus far, no studies have examined brain activity or connectivity in lonely individuals that have received successful interventions. This would certainly be an interesting research topic, and we speculate that since interventions that are aimed at interrupting the negative thought cycle are more successful, this may be associated with increased cognitive control and thus an increase in activity within frontoparietal and/or cingulo-opercule regions.

Not only are we physically separated from much of our social support, but many people are also dealing with the loss of a loved one due to COVID-19. This shared global experience has emphasized the need for continued research on the neural underpinnings not only of loneliness, but of loss and grief processing as well. Like loneliness, loss of a loved one is associated with altered activity in similar brain regions, leading to alterations in motivation and stress reactivity. Although loneliness is a common experience for grieving individuals, research using humans and animal models suggest that the loneliness state during this period is unique from the experience of loneliness in the absence of loss. This is likely due to the structural and neurochemical changes that occur during the formation (and subsequent maintenance) of a strong bond/relationship that appears to be more rewarding, and therefore harder to get over, than isolation from peers. Further research using the prairie vole model of social loss appears to be aimed toward understanding the natural degradation of relationship-related neural pathways, while additional work should assess the features of an individual or a relationship that makes it harder to move on from some compared to others. Investigating loneliness, isolation, and loss through various lenses could allow us to find a new approach to improving the quality of life for those experiencing such conditions.

EV and AS wrote and edited all versions of this manuscript. Both authors contributed to the article and approved the submitted version.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.