The Importance of Social Behavior in Nonhuman Primate Studies of Aging: A mini-review

Abstract

Social behavior plays an important role in supporting both psychological and physical health across the lifespan. People’s social lives change as they age, and the nature of these changes differ based on whether people are on healthy aging trajectories or are experiencing neurodegenerative diseases that cause dementia, such as Alzheimer’s disease and Parkinson’s disease. Nonhuman primate models of aging have provided a base of knowledge comparing aging trajectories in health and disease, but these studies rarely emphasize social behavior changes as a consequence of the aging process. What data exist hold particular value, as negative effects of disease and aging on social behavior are likely to have disproportionate impacts on quality of life. In this mini review, we examine the literature on nonhuman primate models of aging with a focus on social behavior, in the context of both health and disease. We propose that adopting a greater focus on social behavior outcomes in nonhuman primates will improve our understanding of the intersection of health, aging and sociality in humans.

1. Introduction

Understanding what confers a good life from womb-to-tomb is becoming increasingly important as the global population ages, in concert with rising rates of psychiatric and neurodegenerative diseases (Alzheimer’s Association, 2021; Collins et al., 2011). Accumulating evidence demonstrates that having social relationships, generally, and high-quality social relationships, specifically, is particularly vital for promoting well-being across the lifespan. Strong, supportive social relationships can prolong lifespan (Holt-Lunstad et al., 2010; House et al., 1988) while loneliness and social isolation literally kill (Cacioppo & Cacioppo, 2018; Holt-Lunstad et al., 2015). While social relationships are important across the whole lifespan, they seem to be particularly important in older age where they buffer deleterious effects of biological aging (reviewed in Courtin & Knapp, 2017; Luo et al., 2012). Patterns of social relationships seem to change with age, with people prioritizing close social relationships and often reducing their number of social connections. As we will review here, the mechanisms of processing social information – including attention and valuation of social stimuli – undergo significant changes in older adults both who are healthy and those who are experiencing dementia caused by neurodegenerative diseases.

While a few large longitudinal studies of human psychological aging exist (e.g., English Longitudinal Study of Aging: Banks et al., 2023; Midlife in the US: Radler, 2014; Framingham Heart Study: Mahmood et al., 2014), studies in humans are fundamentally limited by the lack of environmental and experimental control across the entire lifespan (i.e., so much of life that either cannot be or is not manipulated or measured) and the inability to carry out causal manipulations to understand mechanisms. As a result, nonhuman animal models (hereafter, animal models) are becoming increasingly important for understanding social aging and ultimately developing treatments and interventions to promote healthspan (physical health into old age) and mindspan (psychological health into old age) (Luo et al., 2012). Of the animal models, nonhuman primates (NHPs) are particularly valuable models for social aging trajectories because they share key features of social life and neurobiological structures with humans that many other species, including most rodents, do not (Phillips et al., 2014). These social features include living in complex social groups that include close kin and non-kin relationships and prolonged early development spent with the mother among others (Phillips et al., 2014). Neurobiologically, most NHP species have highly developed frontal and prefrontal cortices (Preuss & Wise, 2022; Wise, 2008) which are particularly vulnerable in aging (Morrison & Baxter, 2012; Upright & Baxter, 2021). Moreover, NHPs have comparatively long lifespans compared to rodents making them more suitable for translational research on human aging. The longest-recorded lifespans for laboratory mouse and rats are approximately 4 years, whereas maximum NHP lifespans range from 17 years (giant mouse lemur) to 68 years (chimpanzee) (AnAge Database; Tacutu et al., 2018). Here, we make the case that NHPs are ideal models of human social aging and may shed light on how social processing changes in age-related diseases, like neurodegenerative diseases.

2. Studying social behavior in aging is important as it impacts our health and well-being, especially in aging with neurodegenerative diseases.

2.1. Social behavior in healthy aging.

One of the hallmark findings in aging science that has been revealed over the last few decades is that social life is not stable across the lifespan and changes significantly as people age. People’s social relationships and social interactions shift as they transition from middle adulthood into old age in several ways. While the size of people’s social networks increase up to young adulthood (e.g., 25 years old), social network sizes decline steadily in mid-to-later adulthood until around age 60, then decline more slowly after that into old age (Wrzus et al., 2013). Moreover, people invest more energy in fewer, close relationships as they transition into old age, especially focusing on relationships that have positive and rewarding interactions (English & Carstensen, 2014). This refinement of social support is intertwined with what has been termed the “positivity effect” – in later life, older adults prioritize positive information and/or deprioritize negative information (Charles et al., 2003; Isaacowitz, 2022 for a review). Such positivity effects most consistently apply to processing of affective information (i.e., stimuli that convey or are imbued with valenced or hedonic information, e.g., faces depicting anger, faces depicting happiness) rather than affective experience (e.g., feelings of pleasantness/unpleasantness; emotional experiences like anger, happiness, sadness, etc.) (see Isaacowitz, 2022 for a discussion). Support has been found for the positivity bias in attention, memory, and face processing (reviewed in Carstensen & DeLiema, 2018).

Of note, in the human aging literature, social and affective information processing are often confounded, in part because both social information and affective information are self-relevant and prioritized in information processing and in part because social stimuli (like faces) often explicitly have affective content (e.g., facial behaviors related to emotion). One working hypothesis is that all social information is actually affective because even ostensibly “neutral” social information captures attention and is prioritized in psychological processes (Bliss-Moreau, 2017; Bliss-Moreau et al., 2011, 2021; Williams & Bliss-Moreau, 2016); as a result, here, we consider social and affective processing together and not as mutually exclusive.

Together, the two major socioaffective life changes in old age – winnowing of social relationships and prioritization of positive information – may be why some studies demonstrate that people’s emotional lives become more positive with advanced age (English & Carstensen, 2014). The positive emotional trajectory in aging seems to hinge on having supportive social relationships, as health and well-being differ vastly for people who are socially isolated and/or lonely. Social integration is known to impact longevity robustly and positively (Holt-Lunstad et al., 2010), and the absence of social relationships has deleterious effects. Hawkley and Cacioppo (2007) proposed that the negative health implications of loneliness are greater in older age. While a meta-analysis of cross-sectional studies found that social isolation was the most predictive of death below the age of 65 (Holt-Lunstad et al., 2015), there is strong support for Hawkley and Cacioppo’s proposal. A review of the literature in older adults found nearly every study showed negative effects of isolation or loneliness on psychological and physical health, with greater incidence of depression and poor cardiovascular health being the most widely studied outcomes (Courtin & Knapp, 2017). Longitudinal studies support that loneliness is an important risk factor for morbidity and mortality in old age (Luo et al., 2012; Steptoe et al., 2013). Furthermore, lack of social support, specifically the loss or absence of having a spouse or long-term partner, is a risk factor for developing dementia (Rafnsson et al., 2020). Maintaining positive social relationships is critical to promote health and well-being in old age.

2.2. Models of social behavior changes in healthy aging.

The small number of models have been proposed to explain socioaffective changes in healthy aging differ in whether the changes are based on cognitive shifts, physical shifts (wear and tear on physiological systems), or a combination of both. The model which has captured most attention in this domain suggests that patterns of socioaffective aging can be explained in terms of changes to conscious social motivation, such that recognition of decreasing time left in life leads people to shift their priorities from information or knowledge needs (when they are young) to emotional needs and satisfaction (in older age). This idea is referred to as “Socioemotional Selectivity Theory” (Charles & Carstensen, 2003). Evidence for Socioemotional Selectivity Theory comes not only from studies of aging, but also from studies of people with terminal diseases who also demonstrate similar shifts in social processing (Carstensen & Fredrickson, 1998). That is, changes to social life are thought to be conferred not by age per se but by the realization that one’s time left alive is limited (reviewed in Carstensen, 2021). Socioemotional Selectivity Theory has also been used to try to explain positivity effects by means of shifting regulatory mechanisms related to how affective information is processed (see Isaacowitz, 2022 for a discussion).

While Socioemotional Selectivity Theory has largely dominated the theoretical space on human social and affective aging, alternative hypotheses exist. For example, another perspective referred to as “Maturational Dualism” suggests that changes observed in social and affective processing in older age are directly related to normal biological aging (e.g., weakened musculature, neuropathy of the extremities, stiffening of the vessels, reduced cardiac capacity) that occur via normal wear and tear on the body (Mendes, 2010). These changes in the body result in shifts in how information from the autonomic nervous system, and peripheral nervous system more generally, is processed by the brain which results in changes to social and affective information processing and experience. Such physiological aging may drive the positivity effect found in aging. For example, blunted sympathetic nervous system activity has been found in older adults in response to negative emotions (e.g., anger, fear, sadness) (Levenson et al., 1991). Moreover, experiences of high arousal emotions that are negative (e.g., fear, anger) may be more weakly associated with their internal sensations (e.g., racing heart) in older adults because of these naturally occurring changes in the strength of autonomic signals to the brain, contributing to the positivity effect because of the blunting of negative experience (MacCormack et al., 2021).

Finally, the Strength and Vulnerability Integration model mutually considers age-related changes in cognitive and emotional processing that may lead to improved emotion regulation as well as heightened vulnerability with aging (Charles, 2010). According to this model, older adults should avoid negative situations, including negative social interactions, as they are more vulnerable to the physically deleterious effects of stress compared to younger adults. Under this model the positivity effect emerges from avoidance of negative stimuli and inputs. This model is underspecified in terms of specific patterns and mechanisms.

These three models (Social Selectivity Theory, Maturational Dualism, Strength and Vulnerability Integration Model) share common features and predictions, and thus are not mutually exclusive. Investigations from cross-sectional and longitudinal studies find intertwined support for each. A full report of the evidence supporting each model is beyond the scope of this paper, but excellent reviews exist (Carstensen, 2021; Charles & Carstensen, 2010; Isaacowitz, 2022). These three models attempt to account for patterned changes across the lifespan to socioaffective behavior and experience ultimately in two domains:

1. shifts in social processing that prioritize close social relationships and typically reduce the overall number of social relationships, and

2. affective shifts in attention processing, towards positive information/experiences and/or away from negative information/experiences, and possibly a reduction in the processing of arousal-based information.

Ultimately, each model makes slightly different predictions about the mechanisms that underlie changes in healthy aging. Changes are thought to occur via cognitive processes (e.g., recognition that time left alive is limited), affective processes (e.g., stress reduction), and/or processes that impact the function of the nervous system which changes both central inputs from the autonomic/peripheral nervous systems and representations in the central nervous system. The relationship between these possible underlying causes is unclear and could be directional, reciprocal, or unrelated/independent. Without consensus on the mechanisms for age-related changes in social behavior in healthy aging, it remains difficult to describe how the mechanisms are altered in neurodegenerative aging and ultimately difficult to intervene to promote well-being.

2.3. Social behavior in neurodegenerative aging.

While many neurodegenerative diseases are thought to be “cognitive” disorders based on hallmark declines of cognition with disease progression, other features of psychological life are impacted by these disease processes as well. Neurodegenerative diseases are a category of disease where neurons lose function (synaptic phase) and eventually die (atrophy///). In the synaptic phase there are structural and functional changes to individual neurons because of disease specific biochemical processes, followed by a cascade of more significant cellular and ultimately structural atrophy. The diseases vary in their specific cellular and structural pathology, and thus the mechanism of potential intervention. The primary pathologies in Alzheimer’s disease (AD) are amyloid deposits, including plaques, and tau pathologies, including tangles (Braak & Braak, 1991). In behavioral variant frontotemporal dementia (bvFTD), patients present with atrophy primarily in the frontal and temporal lobes of the cortex (Hodges, 2007). Parkinson’s disease (PD) is characterized by a progressive loss of dopamine neurons in the basal ganglia (Walker et al., 2019). We will focus on three neurodegenerative diseases: AD, bvFTD, and PD with dementia, with their behavioral, psychological, and cognitive symptoms summarized in Figure 1. AD and bvFTD both fall under the dementia umbrella and are characterized by cognitive and memory impairments that interfere with daily life. Approximately 30 to 40% of patients with PD also develop dementia and are diagnosed with “Parkinson’s disease with dementia” and the incidence of this diagnosis increases with duration of time a patient has a Parkinson’s diagnosis (Goetz et al., 2009; Bock & Tanner, 2022). These patients develop cognitive and behavioral symptoms more than one year after the onset of well-known motor symptoms (tremors, stiffness, etc.) (Walker et al., 2019). People with dementia, including the above three diagnoses, often have prominent changes in social behavior and affect, and these non-cognitive symptoms are among the most debilitating. Despite the biological differences among these neurodegenerative diseases, there is great overlap in symptom presentation. In fact, it is common for diagnoses to be confused and often patients receive erroneous diagnoses early on. Moreover, there is broad overlap in behavioral and psychological symptoms, suggesting that a common feature among these neurodegenerative diseases is degrading of neural systems that share core features of how socioaffective life is impacted.

FIGURE 1.

Summary of changes in cognition, social life, and affect in healthy aging or neurodegenerative diseases. Prominent and/or unique symptoms are highlighted for each neurodegenerative disease. (AD = Alzheimer’s disease; bvFTD = Behavioral variant frontotemporal dementia; PDD = Parkinson’s disease with dementia).

Greater than 90% of people with dementia caused by neurodegenerative diseases are estimated to be impacted by non-cognitive symptoms (Kales et al., 2015; Laganà et al., 2022), often referred to as “behavioral and psychological symptoms” (Finkel, 2000). The category includes symptoms of mood disorders, like depression and anxiety, as well as agitation, aggression, delusions, hallucinations, apathy, disinhibition, and euphoria. Non-cognitive symptoms in this category also include what is referred to in the literature as alterations in “social cognition” including poor social perception, impaired theory of mind and meta-cognition, reduced empathy and emotional processing, and abnormal social behavior (reviewed in Desmarais et al., 2018; Elamin et al., 2012). Various neurodegenerative diseases differ in the exact presentation of these non-cognitive dementia symptoms, as well as the time course of presentation relative to cognitive symptoms.

Behavioral and psychological symptoms are most prominent in bvFTD and include socially inappropriate behavior such as loss of self-control, loss of empathy, inappropriate affect, disinhibition, reduced perception of social cues, and impaired theory of mind (reviewed in Desmarais et al, 2018). A longitudinal study found that symptoms related to mood disorders (depression, anxiety) presented first for people who were eventually diagnosed with bvFTD and AD, and other behavioral and psychological symptoms presented later (Laganà et al., 2022). For both bvFTD and AD, behavioral and psychological symptoms often present well before cognitive impairments (Sperling et al., 2011). Mild behavioral impairment (MBI) is a recent diagnostic contrast that has been proposed by the International Alzheimer’s Disease Association to identify prodromal stages of Alzheimer’s disease using non-cognitive neuropsychiatric and behavioral symptoms (Ismail et al., 2016). Behavioral and psychological symptoms are less prominent in PD with dementia, and in contrast to bvFTD and AD, these appear later in the disease course (reviewed in Desmarais et al., 2018). People newly diagnosed with PD performed similarly to controls on social cognitive tasks (Bodden et al., 2010; Péron et al., 2009). However, it is important to note that dementia is only diagnosed with PD if behavioral and/or cognitive symptoms emerge greater than one year after motor symptoms. Regardless, while the exact presentation varies across dementia types, disruptive socioaffective symptoms are common across neurodegenerative diseases.

In healthy aging, shifts in social processing that prioritize close relationships and typically result in reduced social network sizes. In aging with neurodegenerative disease, the evidence for this is mixed, and the directionality of any effect is not clear. Small social network size, sometimes considered a proxy for reduced social support, has been evaluated primarily as a risk-factor in developing neurodegenerative disease. For example, the English Longitudinal Study of Aging, a robust longitudinal study carried out in England, found that being married and having close relationships were each independently related to a reduced dementia risk (Rafnsson et al., 2020). Moreover, social support may be protective in disease progression, and Bennett et al. (2006) found in a longitudinal study of AD patients that social network size modified the relationship between cognitive function in life and post-mortem neuropathology such that people with larger network sizes retained higher cognitive function despite high levels of pathology. In these examples, it remains difficult to determine directionality and is further complicated in diseases like bvFTD and AD where behavioral and psychological symptoms occur in prodromal stages of disease. Therefore, social disruptions may be a sign of impending neurodegenerative diagnosis or may be a risk factor that precedes disease onset, or both. The challenge of directionality is exemplified further in another report from the English Longitudinal Study of Aging where people who developed dementia showed decreased social engagement starting before diagnosis whereas healthy controls showed no decline (Hackett et al., 2019).

Another challenge faced in evaluating how social life changes as a consequence of neurodegenerative diseases is understanding the bidirectional interactions between patients and their social partners, who often are their caregivers. For example, seeking support in a nursing home or skilled care facility can exacerbate reduced social engagement, as dementia patients have been found to have less social contact after admission compared to people in nursing homes without dementia (Port et al., 2001). This change could be attributed to reduced social interest by the patient, or social distancing by family, friends, or other caregivers. There is robust evidence that dementia strains relationships and studies show detriments for caregiver well-being, including mental and physical health (reviewed in Feast et al., 2016; Schulz, 2020). Institutionalization rates are also higher among patients who have more behavioral and psychological symptoms (Steele et al., 1990). PD with dementia seems to be an exception, where the interpersonal implications of social behavior symptoms are less clear (discussed in Desmarais et al., 2018). The fact that behavioral and psychological symptoms appear later in disease progression in PD with dementia may also diminish their impact on social relationships and caregivers. Taken together, some studies suggest a similar change in social life between people aging on a healthy vs. neurodegenerative trajectory, such that social lives shrink, and dementia patients may have fewer, more strained relationships, stemming from the socioaffective symptoms that come with neurodegenerative diseases.

Changes in social life in the context of dementia may be the product of changes to socioaffective information processing, namely a reduced ability to recognize negative emotions and/or facial behaviors depicting emotions, which shows some overlap with positivity effects found in healthy aging adults. The exact nature and extent of these impairments vary across disease types. People with bvFTD have poor emotion processing, especially for negative emotions (reviewed in Kumfor & Piguet, 2012). For example, patients with bvFTD were found to have impaired ability to recognize negative facial emotions, including anger, disgust and fear, compared to people with AD (Fernandez-Duque & Black, 2005). However, there is evidence that patients with a variant of FTD (semantic primary progressive aphasia) were appropriately able to categorize facial expressions as positively or negatively valenced, suggesting that emotion perception but not affect perception appear to be compromised in FTD (Lindquist et al., 2014). The connection between emotion recognition and physiological response may be interrupted in bvFTD as another study found patients to have a dampened psychophysiological response (e.g., facial EMG, skin conductance) to positive, negative, or neutral video clips overall (Kumfor et al., 2019).

A systematic review of AD literature found more mixed results in emotion recognition abilities, but happiness was most often identified correctly and negative emotions (e.g., anger, sadness) were least often identified correctly across studies (Torres Mendonça De Melo Fádel et al., 2019). Therefore, patterns among AD patients are similar but less prominent compared to bvFTD. Recognition of emotion in faces is also impaired in PD, with more studies pointing to deficits in recognizing negative emotions (reviewed in Argaud et al., 2018). Together, these findings may support a positivity bias in socioaffective processing for aging with neurodegenerative disease, especially with broad findings of reduced ability to process or recognize negative emotions. However, these findings are rarely discussed as such in the literature (except see Gkinopoulos et al., 2014). As a result, it is hard to say that patients with neurodegenerative disease have a positivity effect when there are major underlying affective shifts that lead to the other behavioral and psychological symptoms such as depression, anxiety, and other psychiatric symptoms. This points to the fact that underlying mechanisms may be changing in a similar way across all aging, regardless of neurodegenerative disease presence, but that the behavioral outcome is vastly different and, and this disconnect should be a focus of future research.

2.4. Models of social behavior changes in neurodegenerative disease.

In the context of neurodegenerative aging, there are few answers and many questions related to how neurodegenerative diseases alter social aging trajectories. One possibility is that neural degeneration alters structures responsible for maintaining social behavior. Alternatively, social impairments may be a result of cognitive impairment that comes with dementia caused by neurodegenerative disease. Finally, it has been proposed that stress and/or the lack of social support may leave people vulnerable to and accelerate disease processes.

Actual neural degeneration and pathology may cause changes in socioaffective behaviors (reviewed in Desmarais et al., 2018). Support for a direct degenerative mechanism fits when the timeline of pathological progression and social behavior changes are matched, such as in bvFTD and AD. In bvFTD, the deficits in socioaffective processing are thought to be related to early atrophy of the frontal and temporal lobes (Lindau et al., 2000). In AD and PD, early neurodegeneration begins in brainstem and other subcortical regions including subcortical areas (e.g., locus coeruleus, dorsal raphe nucleus, substantia nigra and nucleus basalis of Meynert) that mediate behavioral and psychological symptoms (Theofilas et al., 2015). The earliest signs of phosphorylated tau, the component of neurofibrillary tau tangles found in AD, are found in the locus coeruleus, the primary source of noradrenergic projections to the cortex. Evidence has been found of neurofibrillary tau tangles as early as 10 years prior to cognitive decline in AD (Braak et al., 2011). A degeneration mechanism does not fit as clearly for PD with dementia (Walker et al., 2019). Clearly, pathological and degenerative processes in neurodegenerative diseases contribute to behavioral and psychological symptoms, and early degeneration in subcortical regions may underlie socioaffective processes that disrupt and interrupt social life.

An alternative explanation for social behavior changes in neurodegenerative diseases is that social impairments are a direct downstream effect of cognitive impairment. For example, social networks may shrink because patients with dementia may have difficulties socially engaging, such as forgetting to use the phone or attend social events. Similarly, patients with dementia can have a compromised sense of self due to cognitive decline and therefore take less pleasure from social activities (reviewed in Hackett et al., 2019). However, the different timeline of symptom progression with social changes emerging before cognitive changes for dementias suggests that these are not causally connected. Moreover, one study sought to disentangle social and general cognition in aging timelines and found that the two were not related cross-sectionally or longitudinally across time in patients with probable AD (Cosentino et al., 2014).

Other explanations of altered aging trajectories for social behavior in neurodegenerative diseases hinge on the interconnection between social support and stress. Having fewer social partners leaves people without social buffering to stress and can lead to loneliness and/or social isolation (discussed in Fratiglioni et al., 2004; Rafnsson et al., 2020). Fewer social partners also means that if there are poor quality social interactions within close relationships, there may be more negativity, less cognitive stimulation, and greater vulnerability to aging pathology. These explanations are somewhat circular, recognizing the fact that certain social factors (e.g., loneliness, isolation) can lead people to be more vulnerable to age-related diseases including neurodegeneration (reviewed in Drinkwater et al., 2022; Hackett et al., 2019; Rafnsson et al., 2020), but lack of social support in the face of neurodegeneration can accelerate disease processes. This points to the strong bidirectional relationship between psychological well-being and physical health that is critically important in neurodegenerative aging. To truly evaluate the models of social aging in healthy contexts and neurodegenerative disease, experimental research is needed in social animals with high translational value to humans, like NHPs.

3. NHP models are important to evaluate models of social aging and the underlying mechanisms.

3.1. NHP models of human aging.

While well-designed longitudinal and cross-sectional studies in humans reveal possible factors that might influence aging trajectories, they are inherently correlational and it is nearly impossible to thoroughly document and/or account for variation in people’s early environments, access to health care, and lifestyle factors known to influence health. One solution to these challenges is to employ animal models which allow for high fidelity tracking of life histories and standardization of environmental features and medical care. Further, we can capitalize on the similarity of some animals’ behavior, physiology, and neurobiology to humans to ethically test interventions and therapeutics aimed at improving disease outcomes and maintaining high quality of life into older age.

NHPs are widely used models for human health and disease because of the homologies they share with humans in a variety of processes including genetics, physiology, social complexity, affective and cognitive processes, and structural and functional aspects of the brain (Phillips et al., 2014; Ross & Salmon, 2019; Shively et al., 2021). NHP models have several advantages over rodent models with regards to aging research insofar as they naturally recapitulate the biological and psychological process of humans more closely (Phillips et al., 2014). These advantages start in early development as rodents are born altricial creating vastly different early life experiences from those of human and NHPs who are moderately precocial. The extended longevity of NHPs is relevant in the study of aging as well, as it may indicate evolved defenses against the aging process that may be shared between humans and NHPs, but not amongst shorter lived species like rodents (Chiou et al., 2020). Rhesus monkeys, the most common captive NHP, have a median lifespan of 25 years (Walker, 1995) and mature three to four times faster than humans (Higley et al., 1992). Common marmosets, a small-bodied NHP species used increasingly for aging and neuroscience research, have a typical lifespan in captivity of 5 to 13 years (Ross, 2019; Tardif et al., 2011; Nishijima et al., 2012) and mature nine to 12 times faster than humans. Therefore, it is possible for a single researcher to study NHP lifespans in one (e.g., marmosets) or a few (e.g., macaques) conventional grant cycles. Finally, while rodent colonies rely on inbreeding to produce specific strains, NHP research uses outbred subjects which maintains a higher level of inter-individual variability which likely better models human biology (Colman, 2018).

Much like humans, with age, NHPs show changes in their physical appearance and ability (e.g., thinning and graying of hair, skin atrophy, slower gait speed) (Chiou et al., 2020; Colman, 2018; Latimer et al., 2019) and increased incidence of disease (e.g., cancer, cataracts, osteopenia, cardiovascular disease) (Chiou et al., 2020; Colman, 2018). Along with these physical changes, there are changes in the brains of aging NHPs that mirror those observed in aging humans and are associated with worsening cognitive function (reviewed in Freire-Cobo et al., 2021; Upright & Baxter, 2021). These changes include cortical thinning (Koo et al., 2012), degeneration of myelin (Shobin et al., 2017), reduced cell proliferation in the hippocampus (Ngwenya et al., 2015), and synaptic alterations (e.g., synaptic loss, reduction in dendritic spines, loss of multisynaptic boutons) (Morrison & Baxter, 2012; Upright & Baxter, 2021). Age-related cognitive changes in several NHP species are well-documented (e.g., rhesus macaques: Rapp & Amaral, 1992; Herndon et al., 1997; African green monkeys: Cramer et al., 2018; common marmosets: Rothwell et al., 2021; Sadoun et al., 2019; reviewed in: Lacreuse et al., 2020). These parallels in the health effects of age in NHP species display the variable and naturally occurring physical health and cognitive decline outcomes that can be studied using NHP models.

3.2. Natural social aging in NHPs.

A growing body of evidence demonstrates that NHPs exhibit similar patterns of social and affective aging, with the potential for some nuanced differences (reviewed in Machanda & Rosati, 2020) that are not well characterized. First, it does appear that NHPs shift social processing that prioritize close social relationships and typically reduce the overall number of social relationships. Like humans, aged NHPs, compared to younger animals, appear to have smaller social networks composed of potentially more meaningful connections (tufted capuchins: Schino & Pinzaglia, 2018; female rhesus macaques: Corr, 2003; Siracusa et al., 2022; Japanese macaque: Hauser & Tyrrell, 1984; female Barbary macaques: Almeling et al., 2016, 2017; male chimpanzees: Rosati et al., 2020).

Biological sex differences are an important caveat among NHP studies of social behavior (discussed in Machanda & Rosati, 2020 – note that for nonhuman animals we discuss only sex and not gender). Many studies focus on males or females based on their naturalistic social structure, because social structures differ across sex. For example, Negrey et al. (2023) found that older captive female vervet monkeys spent less time in affiliative interactions and more time alone but did not receive less grooming from conspecifics. In the cases where both sexes are examined, different patterns are often found for males than for females. For example, social interest and engagement in aged NHPs has mixed results based on sex. Aged male rhesus macaques showed greater social engagement compared to younger males (Hauser & Tyrrell, 1984) and social interest appears to be maintained in aged female Barbary macaques (Almeling et al., 2016). In contrast, a decline in social engagement was observed for females in long-tailed macaques (Veenema et al., 1997), and captive common marmosets, with females showing longitudinal decline in initiating social interaction as they aged from middle to early old age (Rothwell et al., 2021). NHP models can help clarify mixed evidence about whether social network size and/or the amount of social support is a risk factor and/or consequence of aging since directionality is not clear in human work. Siracusa et al., (2022) is an example of how we can evaluate the models of social aging for healthy human with greater control in NHPs. These authors evaluated several contingencies that may have led to support for social selectivity where it was not appropriate, such as diminished social networks simply due to loss of social partners because of death, or that older individuals were less attractive as social partners to other individuals in the group (i.e., they were not social outcasts). Ultimately, Siracusa et al., (2022) presented compelling evidence that semi-free ranging female rhesus macaques showed longitudinal decreases in social network size, with closer relationships remaining in old age, mirroring patterns of increased social selectivity found in healthy aging humans.

Researchers studying NHPs have also evaluated the positivity effect found using both data from social interactions and studies of socioaffective information processing. For example, negative social interactions (e.g., aggression) and positive social interactions (e.g., grooming) can be quantified and compared across the lifespan. Wild male chimpanzees maintained grooming but decreased in aggression with age, which was interpreted as a positivity bias in social behavior (Rosati et al., 2020). Increases in grooming have also been found in female Japanese macaques (Nakamichi & Shizawa, 2003). In contrast, what evidence exists suggests that there are not changes to negative social interactions with age. Tufted capuchins (Schino & Pinzaglia, 2018) and female Barbary macaques (Almeling et al., 2017) showed no change in aggression with age.

Patterns of socioaffective processing are less well characterized across aging in NHPs. Attentional bias has been studied by showing photographs of socioaffective stimuli like photographs of close or unfamiliar conspecific in positive, negative, or neutral contexts or photos of conspecifics generating facial behaviors thought to convey positive, negative, or neutral affect. Rosati et al. (2018) reported the opposite of a positivity effect in older rhesus monkeys, with visual attention bias for looking at some negative photos (e.g., threatening faces of female stimuli) compared to neutral expressions. There were no age-related differences in looking time when viewing positive (e.g., lipsmack of male stimuli) compared to neutral photos faces. Also, older monkeys looked less at all photos overall compared to younger monkeys. Almeling et al. (2016) found that Barbary macaques looked for longer at photos of conspecifics from close relationships (e.g., “friends”) compared to non-close conspecifics (e.g., “non-friends”), but this did not differ between younger and older monkeys. Both studies (Rosati et al., 2018; Almeling et al., 2016) used visual attention measures derived from video recordings and presented very few stimuli to a fairly large number of monkeys – a fairly different design from those employed when studying humans.

There is evidence that NHPs experience similar valence-based attentional shifts when methods from the human attention literature are used. Using an eye tracking task designed to emulate those used in humans (e.g. Isaacowitz et al., 2006), visual attention patterns that suggest a positivity effect were identified in male and female rhesus macaques (Santistevan, et al., 2022). Middle-aged monkeys looked for longer at faces depicting negative information (e.g., threat face) but aged monkeys showed no such bias. Despite good tools being available to translationally evaluate autonomic nervous system responses (e.g., heart rate, respiratory sinus arrythmia, pre-ejection period, blood pressure) and evidence that NHPs have similar patterns of autonomic physiological responses in response to affective stimuli as humans do (Bliss-Moreau et al., 2013), only one recent study looked at age-related changes in cardiac physiology in response to socioaffective stimuli. Replicating the original report (Bliss-Moreau et al., 2013), middle-aged monkeys’ cardiac parasympathetic responses tracked with the affective valence of the stimuli they viewed (30-second movies that varied from very negative to very positive), but aged monkeys’ cardiac responses were not impacted by valence and arousal content to the same degree – that is, their ANS reactivity was blunted (Santistevan et al., 2022). Baseline parasympathetic activity was consistent across age groups and there was no evidence of age-related differences in sympathetic function. In contrast to the lack of age-related impacts on baseline ANS function, Shively et al., (2020) found that female cynomolgus macaques showed longitudinal changes in autonomic nervous system activity, with increased sympathetic and decreased parasympathetic activity across age. Clearly, further research is needed.

4. NHP models of neurodegenerative aging have untapped potential for the study of social behavior.

4.1. NHP models of neurodegenerative disease.

NHPs are becoming an increasingly popular model for neurodegenerative disease like AD and PD both because of the failure of rodent models to produce successful treatments and because NHP models exhibit higher cognitive functions that are similar to humans (De Felice & Munoz, 2016). Some evidence exists suggesting naturally occurring brain changes in aging NHPs mirror changes observed in humans with neurodegenerative diseases (reviewed in Freire-Cobo et al., 2021), even if the full neural phenotype is not recapitulated. As in humans with AD, older great apes and older monkeys develop both amyloid deposits and plaques and tau pathologies (chimpanzees: Elder et al., 2017; gorillas: Perez et al., 2013; orangutans: Gearing et al., 1997; rhesus macaques: Arnsten et al., 2019; Cai et al., 2010; cynomolgus macaques: Oikawa et al., 2010; marmosets: Freire-Cobo et al., 2023; Geula et al., 2002; Rodriguez-Callejas et al., 2016; Ross & Salmon, 2019) and behaviorally demonstrate cognitive decline with older age (rhesus macaques: Hara et al., 2012; Peters et al., 1996; marmosets: Rothwell, et al., 2021; Rothwell, et al., 2021; Sadoun et al., 2019). Aging rhesus monkeys, like aging humans show increased alpha-synuclein proteins in neurons in the substantia nigra and decreased tyrosine hydroxylase (an enzyme involved in dopamine synthesis) which may contribute to degeneration of dopaminergic neurons characteristic of PD (Chu & Kordower, 2007). Additionally in a case study of a cynomolgus macaque with spontaneous PD, both physiological (e.g., dopaminergic neuron loss in the substantia nigra pars compacta and evidence of gliosis) and behavioral (e.g., dysregulated balance and poor gross motor skills) indicators aligned with human pathology (Li et al., 2021). These naturally occurring changes in the brain associated with aging demonstrate the potential for studying behavioral shifts in animal models because of homologous pathology to humans.

NHP models also exist which capitalize on the ability to induce disease-like neuropathology. Experimentally manipulated models of AD and PD in NHPs provide a unique context in which to study both neuropathological and behavioral effects of disease state. In AD research, manipulated models have been developed through exogenously introducing amyloid beta (e.g., Beckman et al., 2019; Forny-Germano et al., 2014; Leung et al., 2011; Philippens et al., 2016; Wakeman et al., 2022) or tau (Beckman et al., 2021) into the brain and assessing the degree to which neuropathology is observed. A model of early-onset AD has also been developed in marmosets using genetic engineering by knocking in point mutations in the presenilin 1 gene, which causes early-onset AD in humans (Sukoff Rizzo et al., 2023). PD research often involves the administration of 1-methyl-4-phenyl 1,2,3,6-tetrahydropyridine (MPTP) as a neurotoxin to destroy dopaminergic neurons and initiate motor symptoms. Results from this kind of research have demonstrated successful recapitulation of both expected neuropathology and some behavior in NHP models of disease. Physiological outcomes of MPTP models include measures of dopaminergic neuron loss (Blesa et al., 2012; Masilamoni & Smith, 2018), and central and peripheral inflammation (Joers et al., 2020).

4.2. Behavior in NHP models of neurodegenerative disease.

While several species have been used to understand elements of AD and PD (reviewed in Freire-Cobo et al., 2021) including vervet monkeys (Chen et al., 2018; Frye et al., 2021), marmoset monkeys (PD: reviewed in Eslamboli, 2005), macaques (PD: Bezard et al., 1997; Joers et al., 2020; Potts et al., 2014), and chimpanzees (Heuer et al., 2012), these examinations largely lack a focus on behavior and when they do incorporate behavior it is typically primarily focused on cognition. AD research in NHPs often focus on neuropathologies more so than behavioral outcomes and experimentally induced AD studies largely lack behavioral complements to their histological work leaving open questions about the impacts that these documented shifts in brain structure have on functional outcomes. Common behavioral outcomes of NHP models of PD include cognitive assessments (e.g., measures of cognitive flexibility, working memory and executive functioning, Joers et al., 2020; Vezoli et al., 2011), measurements of motor function (e.g., observer ratings for freezing, tremors, rigidity, posture, bradykinesia, and ability to move food, Shi et al., 2020; Vezoli et al., 2011), and rest-activity cycles (measured with infrared movement detectors, macaques, Vezoli et al., 2011).

Though in the human literature it is apparent that shifts in social cognition and social behavior have the potential to impact disease progression, little research in NHPs has focused on this area. The authors are not aware of any social behavior outcomes published on experimental NHP models of AD. One such study that did include an analysis of social behavior in a PD model of long-tailed macaques identified changes in social behavior that occurred prior to any motor- or cognitive-related symptoms (Durand et al., 2015), highlighting the importance of subtle changes to social behavior in potentially identifying early pathological stages. These findings are bolstered by a second group that found male marmoset monkeys displayed decreased attention toward their female social partner after treatment with MPTP (part of a PD model) compared to before (Phillips et al., 2017). Consistency across NHP models is promising for clarifying mixed evidence for the time and nature of social behavior change in humans with PD with dementia. However, these two small studies in isolation are not enough to draw broad conclusions and comparisons to humans.

4.3. NHPs studies of social behavior in healthy & neurodegenerative aging: Promises & Pitfalls

Expanding the study of social aging in NHPs across species and contexts will give insights into social aging across all primates, including humans, in both health and disease. NHP studies are well-suited to evaluate and refine the models of socioaffective changes in healthy aging. The three models discussed here share common features, they vary in their dependence on cognitive awareness of time horizons and aging from none (Maturational Dualism) to complete dependence (Social Selectivity Theory) and something in between (SAVI). In all likelihood NHPs do not have the capacity to sense their time horizons in the same way that humans do, but nonetheless there is evidence similar shifts in socioaffective behavior including social selectivity. This lends support for Maturational Dualism, which centers on biological and physiological declines that are common across all primates. Recently, NHP and other animal researchers have proposed means by which evolution would drive social selectivity and positivity effects without cognitive awareness of aging to explain this commonality in humans and NHPs. Siracusa et al., (2022) outlines three evolutionary drivers that may lead to these patterns across mammals, including primates, including that 1) social behavior changes are due directly to body and brain aging that may impact physical or cognitive traits; 2) social behavior changes that are adaptive in light of body and brain aging; 3) experience accumulated across aging provides better social skills and therefore positive changes. Gonzales et al. (2023) proposes for NHP specifically that being socially selective conveys benefits to individuals as they navigate their environments while facing physical and cognitive declines. Here, putting humans in evolutionary context with NHPs highlights refinement of and challenge to human-focused models of social aging.

Studying NHPs can also help evaluate the models that seek to describe how socioaffective processes are altered in neurodegenerative aging. Longitudinal studies of NHP that concurrently track socioaffective behavior and aging outcomes can clarify whether social factors like social isolation is causally related to healthspan and/or lifespan (e.g., Tung et al., 2023). Likewise, controlled longitudinal designs or mediation analyses (e.g., Negrey et al, 2023) can tackle the degree to which social impairments are a downstream result of cognitive impairment. It is critical to incorporate socioaffective assessments for induced models of neurodegeneration and pathology in NHP, such as infusing amyloid-beta oligomers into the brain, seeding the brain with phosphorylated tau, or administering an agent that would experimentally disrupt normal neurobiology (e.g., MPTP). In these designs social behavior and socioaffective processing can be evaluated prior to induction and again after disease pathology is initiated. Induced NHPs should be compared to control animals to distinguish between naturally occurring changes with aging vs. altered changes in pathological contexts. Socioaffective assessments could include attention to social stimuli, behavioral or physiological response to affective stimuli, and changes to normative social interactions after treatment administration, which can be tracked with common cognitive or histological endpoints. Longitudinal tracking of socioaffective measures in concert with cognitive measures and in vivo assessments of neurodegeneration (e.g., amyloid beta PET) will help determine whether behavioral and psychological symptoms are risk factors for neurodegeneration, prodromal symptoms, or a combination of both.

There is no single NHP model that is best to translationally understand human aging, but the breadth of NHP species and contexts captures important similarities and differences. For example, shorter-lived NHPs like mouse lemurs and marmosets (5 −10 yrs) allow us to study aging at an accelerated rate, such as capturing entire adulthood in detail, but may sacrifice some features of neural complexity. Theoretically, there are major advantages to studying non-human great apes (i.e., chimpanzees, bonobos, gorillas, orangutans) is that their lifespans are much more similar to humans, indeed often decades longer than monkeys, and they have much larger brains and sophisticated cognitive and social abilities. However, great apes can no longer be used for experimental research, limiting the available experimental approaches. An important caveat to NHP research on the current topic is that most of the studies evaluating social aging in healthy contexts represent NHP species and groups that have been studied in wild or semi-free ranging populations, with limitations for biological sampling including studying the brain directly. In contrast, experimental studies of aging related to neurodegenerative diseases have been done in captivity, with limitations that laboratory environments will never be able to recapitulate the natural physical and social environments for each species. Therefore, direct comparison across NHP species is limited and a body of literature will need several species, approaches, and contexts to gain a holistic view of NHP socioaffective aging for comparison with humans.

5. Summary & Conclusions.

Social behavior is the cornerstone of garnering and maintaining positive and supportive social relationships, and these social relationships are a critical component to ensuring physical health and psychological well-being (including affective/emotional processing), especially into old age. Healthspan and mindspan are increasingly recognized as important to prolonging human aging, because quality of life is intricately tied to psychological well-being and the absence of physical disease. Ultimately, targeted interventions to promote a good life and treatments for neurodegenerative diseases should include consideration of socioaffective processes and their development necessitates understanding the mechanisms by which socioaffective processes change in lifespan health and disease. In humans it remains unclear what the cause of social behavior changes are in healthy aging, and theoretical models have been proposed that are overlapping and often difficult to falsify, perpetuating the challenge of explaining debilitating socioaffective changes in neurodegenerative diseases. This is where the power of NHP models lie, as they are excellent for studying aging given practical timelines conferred by shorter lifespans and greater control over social and environmental influences on aging trajectories.

Critically, many NHPs experience similar changes in central nervous system and peripheral biology during aging. Likewise, it is also well-established that NHPs sophisticated social lives make them far better translational models for human health and disease than rodents.

When NHP models are developed, evaluating behavioral features of the models often trails behind establishing the neurological (or other biological) phenotype of the diseases, and clearly models of neurodegenerative diseases are no different. Ongoing research programs at a number of institutions are focusing heavily on cognitive outcomes of these diseases, in part because of the robust history of studying age-related changes to cognition in NHPs (Gallagher & Rapp, 1997) and in part because dementia is thought to largely be a “cognitive” disease. Likewise, models of PD focus on creating and evaluating motor symptoms and ignore the high prevalence of dementia in PD which include cognitive and behavioral symptoms. Given the social and affective symptoms that arise in neurodegenerative diseases, NHP models are likely to be an important tool for both understanding the mechanisms by which behavioral symptoms emerge, their timing relative to other hallmark symptoms, and ultimately developing treatments and interventions for those symptoms. That is, NHP models of neurodegenerative disease would be most powerful if they evaluate the full array of behavioral symptoms and are not limited to solely cognition (for AD and other memory-related dementias) or movement (for PD) symptoms. As induced models of neurodegeneration in NHPs are on the rise, this is a critical time to focus on experiments to study the causal relationship between early shifts in social behavior and early neuropathology.

Highlights.

• Social and emotional lives improve for healthy humans as they advance in old age.

• Neurodegenerative diseases that cause dementia results in detrimental changes in both social and emotional lives.

• Nonhuman primates are ideal to test causal models of social changes in both healthy and disease state aging.

• Studies of both human and nonhuman primate models largely overlook social and affective outcomes.