Recognising depression in non-human primates: a narrative review of reported signs of depression

Depressed mood (S1)

Depressed mood (i.e., being sad, hopeless or discouraged) is one of the two core symptoms of depression (American Psychiatric Association, 2013), being reported by 97% of patients with depression (Bryan et al., 2008). If not self-reported, depressed mood can be assessed by reading the person’s facial expression and demeanour, for example, by observing tears (American Psychiatric Association, 2013). However, in NHPs, there is no evidence that certain facial expressions or demeanors indicate a depressed mood (Kret et al., 2020). Still, two behaviours accompanying a depressed mood can be observed in both humans and NHPs: judgement biases and attentional biases.

Depressed humans are more likely to interpret neutral cues, such as neutral facial expressions, as negative (Everaert, Podina & Koster, 2017; Bourke, Douglas & Porter, 2010, Maniglio et al., 2014; Van Vleet et al., 2019). These biases can be assessed by judgement bias tests (JBTs; Lagisz et al., 2020; Mendl et al., 2009). In JBTs it is assumed that judgement and attention are influenced by emotional state (Coleman & Pierre, 2014). JBTs have been successfully used in a variety of vertebrates (Bateson & Nettle, 2015; Douglas et al., 2012; McCoy et al., 2019) and even in insects (Baracchi, Lihoreau & Giurfa, 2017). Non-human animals can be trained to recognise positive and negative cues, and their level of optimism or pessimism can be measured subsequently (Mendl et al., 2009). For example, JBTs have been used to assess the judgements of rhesus macaques (Macaca mulatta) as a function of husbandry conditions (Bethell et al., 2012a). Although the overall effect sizes of results from JBTs are small, manipulations generally had the expected effect, validating JBTs as an indicator of affect in non-human animals (Lagisz et al., 2020). A downside of JBTs is that they require substantial training (e.g., Bateson & Nettle, 2015). Therefore, while JBTs validly measure depressed mood, applying the task is not feasible.

Another cognitive bias in depressed humans is an increased attention towards negative cues, such as sad faces (Keller et al., 2019; Peckham, McHugh & Otto, 2010). Several tasks with varying complexity measure attentional biases in NHPs (Bethell & Pfefferle, 2023; Crump, Arnott & Bethell, 2018): preferential looking time tasks, dot-probe tasks, and emotional Stroop tasks.

In a preferential looking task (Winters, Dubuc & Higham, 2015), an NHP watches a side by side on-screen presentation of a neutral and a negative stimulus (e.g., a neutral and an aggressive facial expression, Bethell et al., 2012b). Depressed humans tend to look more at the negative stimulus after a stressful event. Interestingly, rhesus macaques showed the opposite pattern after a veterinary check (Bethell et al., 2012b), possibly indicating avoidance of threatening behaviour. Moreover, although stressful, a veterinary check is not expected to induce depression. In contrast, sheep (Ovis aries) that were pharmacologically treated to be in a depression-like state did prefer to look towards a threatening compared to a positive stimulus (Monk et al., 2018). Because of the contrasting evidence, the preferential looking task is not (yet) valid to study depressed mood in NHPs.

Depressed humans also show attentional biases in dot-probe tasks (Peckham, McHugh & Otto, 2010), in which a neutral and negative stimulus are displayed side by side on a screen. When the images disappear, one stimulus is replaced by the probe, typically consisting of two dots. Humans are instructed to respond to this dot-probe as quickly as possible (e.g., by touching it on the screen). The latency to respond the dot-probe is lower when the subject’s attention was on the stimulus that was replaced by the dot-probe, thereby revealing biases in attention (Van Rooijen, Ploeger & Kret, 2017). Dot-probe tasks were successfully adapted for rhesus monkeys (Parr et al., 2013), bonobos (Pan paniscus: Kret et al., 2016), chimpanzees (Wilson & Tomonaga, 2018), and long-tailed macaques (cynomolgus monkeys, Macaca fascicularis; Cassidy et al., 2023). NHPs need to be familiarised with the set-up and trained to respond to the dot-probe. In bonobos, this required a 15-week training programme (Kret et al., 2016). However, more work is needed to validate whether increased attention to a specific stimulus is a sign of depression in NHPs. Furthermore, human performance in dot-probe tasks is also influenced by anxiety (Van Rooijen, Ploeger & Kret, 2017). This makes it difficult to disentangle effects of the two disorders. In short, dot-probe tasks might elicit changes in attention resulting from depression, but because of validity issues and the requirement of extensive training, this task is not very suitable for measuring depressed mood in NHPs.

A third way of measuring attentional biases is with an emotional Stroop task (Stroop, 1935), in which depressed and healthy humans perform differently (Joyal et al., 2019; Peckham, McHugh & Otto, 2010). The emotional Stroop task is based on the notion that emotion distracts an individual from performing a task, such as naming the colour that a word is written in Crump, Arnott & Bethell (2018). Participants are instructed to name this colour as quickly and accurately as possible. When the content of the words is negative (e.g., pain), the response time increases. The emotional Stroop task has been adapted for multiple NHP species (Allritz, Call & Borkenau, 2016; Hopper et al., 2021; Laméris et al., 2022). In short, NHPs are trained to choose a picture over another adjacent picture based on differently coloured picture borders on a touch screen. The content of the picture can reflect positive, neutral, or negative information. The reaction time is then compared between the three content categories (Allritz, Call & Borkenau, 2016). Despite the adaptation for NHPs, no studies have yet used the task as an indicator of depressed mood. A major limitation of the emotional Stroop task is the inability to rule out that the differences in reaction time reflect biases in motor actions (Crump, Arnott & Bethell, 2018). Moreover, training NHPs to participate in the emotional Stroop task can take up to thousands of trials, if successful (Allritz, Call & Borkenau, 2016; Hopper et al., 2021). For these reasons, we do not consider the emotional Stroop task a valid or feasible measure of depressed mood in NHPs.

Altogether, attentional bias tasks are not valid measures for depressed mood in NHPs. Judgment bias tasks are suitable, although these are not easily implementable because training is required.

Anhedonia (S2)

The DSM-5 states that a patient suffering from anhedonia, reported in 71% of depressed humans (Cao et al., 2019), does not find pleasure in previously rewarding activities (American Psychiatric Association, 2013). There is a distinction between the experience of rewarding stimuli (consummatory anhedonia) and the motivation to pursue rewarding stimuli (motivational anhedonia) (Treadway & Zald, 2011).

Anhedonia is observable, as it may result in social withdrawal and reduction of sexual interest (American Psychiatric Association, 2013). Behaviours like social play, social interaction with offspring, and sexual behaviour are rewarding for both humans and other animals (Paredes, 2009, 2014; Trezza, Campolongo & Vanderschuren, 2011; Vanderschuren, Achterberg & Trezza, 2016) and can be observed with a relatively low labour investment. As the systems and mechanisms of sociality are homologous across mammals (Crockford et al., 2013; Massen, Sterck & de Vos, 2010), motivational anhedonia may result in a decrease in social and sexual behaviours in NHPs and humans alike. Still, social and sexual behaviours can change when the group composition changes (Caselli et al., 2023; Ryan & Hauber, 2016; Schel et al., 2013). However, if this alternative explanation is ruled out, diminished social behaviours can be a valid indicator of motivational anhedonia in NHPs.

Sugar consumption is another rewarding behaviour in humans and other animals (Hajnal, Smith & Norgren, 2004; Olszewski et al., 2019). Therefore, reduced preference of sweetened water (sucrose) over unsweetened water may indicate decreased reward sensitivity and thus anhedonia in NHPs (Li et al., 2020; Qin et al., 2015; Teng et al., 2021), yet this measure of anhedonia is not undisputed. Sucrose preference decreased in maternally deprived rhesus monkeys compared to a non-deprived control group, but the monkeys also consumed more bitter (quinine) water (Paul, English & Halaris, 2000). Thus, the monkeys did not display anhedonia, but rather became more insensitive to flavours (Paul, English & Halaris, 2000). Moreover, the translational value of sucrose preference in non-humans to anhedonia in humans is still unclear (Markov, 2022). In addition, setting up a sucrose preference test involves a labour investment. Altogether, more research is needed to validate sucrose preference as a measure of anhedonia in NHPs.

Furthermore, humans suffering from depression show less anticipation towards rewards (Rzepa, Fisk & McCabe, 2017; Stringaris et al., 2018). Anticipatory behaviour is defined as a change in behaviour prior to an anticipated rewarding event and can also be measured in non-humans: when animals expect a reward, they present either increased or decreased activity (Krebs et al., 2017; Von Frijtag et al., 2000). Anticipation of rewards represents an appetitive behaviour (Von Frijtag et al., 2000), indicative of reward sensitivity (Kamal et al., 2010; Van der Harst & Spruijt, 2007; Watters, 2014). Although anticipatory behaviour is considered a welfare indicator in various mammalian species (Buchanan-Smith, 2011, Dudink et al., 2006, Vinke, Van Den & Spruijt, 2004), it has not yet been used as an indicator of anhedonia in NHPs. As an anhedonic NHP experiences less pleasure from rewards, it may also have a decreased reward sensitivity (Wooldridge et al., 2021). In this way, anticipatory behaviour can be used to assess anhedonia in NHPs. Because captive animals often learn to associate feeding moments with other stimuli via classical conditioning (e.g., Dickinson & Mackintosh, 1978), anticipatory behaviour can be easily observed in captivity. Thus, although anticipatory behaviour has not yet been used to assess anhedonia in NHPs, we highlight its potential as an indicator.

Overall, social behaviour is the most valid and most feasible way of assessing anhedonia in NHPs. We do not currently regard diminished sucrose preference as evidence of anhedonia and propose anticipatory behaviour as a new measure of anhedonia in NHPs.

Differences in appetite or weight (S3)

According to the DSM-5, depression may decrease or increase appetite, which can impact food intake and consequently, lead to weight change (American Psychiatric Association, 2013). A total of 40% of depressed patients report weight gain, while 30% report weight loss (Caroleo et al., 2019). More than 5% weight change in a month is considered significant in the DSM-5. In NHPs, appetite changes cannot be directly observed, but food intake can be observed without disturbing the animal. Animals can also be weighed, although that requires either capturing the animal or training it to voluntarily cooperate in a weighing procedure (Zijlmans et al., 2021). Alternatively, body condition scoring can be used as a semi-quantitative method to reliably assess body fat and muscle (Clingerman & Summers, 2005). However, this requires the animal to be immobilised, which is highly intrusive. Instead, obesity class can be assessed reliably by observations using the Obesity scale (Berman & Schwartz, 1988). This allows for the determination of obvious weight change without the need to disturb the animal. Similar to the DSM-5, a change in obesity class refers to a weight difference of ≥5% (Zijlmans, Langermans & Sterck, 2019). Motivation for food could also be used as a measure for appetite, but as an animal that is not motivated for food may also be anhedonic, motivation for food is not a truly valid measure for appetite. In short, two valid and feasible ways to assess appetite and weight changes in NHPs exist: observing an animal’s food intake and visually observing weight change. Voluntary weighing is a feasible option for trained individuals.

Insomnia or hypersomnia (S4)

Depression is associated with both insomnia and hypersomnia. Insomnia is estimated to be present in 50–90% of depressed patients, whereas 16–20% of patients report hypersomnia (Tsuno & Ritchie, 2005). In NHPs, sleep patterns can be observed. Observing an animal’s full circadian rhythm provides the most valid indication of sleep patterns, yet is very labour intensive. Alternatively, a valid indication of sleep patterns may be drawn by using systematic short focal samples (Altmann, 1974; Hennessy, Chun & Capitanio, 2017). Moreover, sleep data may be acquired through automated methods (Rushen, Chapinal & de Passilé, 2012; Whitham & Miller, 2016; Green, 2018). For example, non-invasive audio recording devices capture sounds that indicate sleep interruptions (Ayuso et al., 2023). However, these devices have only been used to measure sleep at the group level. Alternatively, individual activity can be measured with accelerometers in captive and wild NHPs (Kooros et al., 2022; Schork et al., 2023). Activity patterns during the night indicate periods of sleeping and waking. However, putting on collars requires handling the animals. Lastly, sleep patterns can be derived from automated methods to measure movement from (infra-red) videos (Green, 2018; Schork et al., 2024). While still not well developed to measure individual sleep and requiring visibility of the sleeping individual, this forms an interesting future avenue to measure sleep non-invasively.

In sum, several methods exist for assessing sleep disturbances in NHPs. These concern, at a group level, nocturnal audio monitoring and automated infra-red recording. At the individual level, non-invasive systematic focal sampling can be used, whereas employing accelerometers is a more invasive method.

Psychomotor agitation or retardation (S5)

The DSM-5 specifically states that psychomotor agitation or retardation must be visible by others, rendering this an observable symptom in humans (American Psychiatric Association, 2013). Psychomotor agitation is present in 10–46% of depressed patients (Sobin & Sackeim, 1997) and its characteristics (difficulties sitting still, pacing, or various ways of fidgeting) can be observed or assessed through self-report (American Psychiatric Association, 2013). In NHPs, psychomotor retardation is sometimes assessed using open field tests (e.g., Li et al., 2020; Teng et al., 2021). More or less locomotion in open field tests may reflect psychomotor changes (Grønli et al., 2005), However, open field tests were not designed to assess psychomotor changes, so their validity is disputed (MacLellan et al., 2021). Alternatively, psychomotor changes may be better reflected by activity in the home environment (MacLellan et al., 2021), which can be measured non-invasively by locomotion frequency and duration (e.g., Li et al., 2020; Teng et al., 2021). Thus, instead of using the disputed open field test, psychomotor agitation is best assessed in NHPs by measuring activity in the animal’s home environment.

The DSM-5 characterises psychomotor retardation with slowed speech, thinking, and body movements, which have to be observable by others (American Psychiatric Association, 2013). Humans with psychomotor retardation exhibit a fixed gaze, slower movement, decreased movement, and a slumped posture (Buyukdura, McClintock & Croarkin, 2011). A total of 62% of patients with depression report psychomotor retardation (Bryan et al., 2008). However, psychomotor retardation can also be a residual effect of past depressive episodes in humans (Gorwood et al., 2014). In NHPs, reduced locomotion frequency and duration is often used as an indicator for psychomotor retardation, either in the home environment or open field (e.g., Li et al., 2020; Teng et al., 2021). As mentioned in the previous paragraph, locomotion changes in the animal’s home environment are a feasible and valid measure of psychomotor retardation, but we cautiously treat the outcomes of open field tests.

Furthermore, the slumped body posture observed in humans with psychomotor retardation shows similarities with a behaviour in NHPs that is called hunching (e.g., Kaufman & Rosenblum, 1967; Lopresti-Goodman, Kameka & Dube, 2012; Shively et al., 1997). Hunching is defined as a “slumped or collapsed body posture”, while having open eyes to distinguish from sleeping (Shively et al., 1997). Hunching is recorded in pigtail macaques (Macaca nemestrina; Kaufman & Rosenblum, 1967), bonnet macaques (Macaca radiata; Rosenblum & Paully, 1984), rhesus macaques (Camus et al., 2014; Hinde, Spencer-Booth & Bruce, 1966), long-tailed macaques (Shively et al., 1997), and in chimpanzees (Goodall, 1986; Lopresti-Goodman, Kameka & Dube, 2012). Because of the great similarity with human posture changes, hunching is a good observable indicator of psychomotor retardation in NHPs.

Altogether, psychomotor agitation and retardation can be assessed non-invasively by recording locomotion frequency and duration. Additionally, hunching can be used as a non-invasive indicator for psychomotor retardation.

Fatigue or loss of energy (S6)

A depressed human patient may report fatigue without physical exertion and find that simple tasks require more time and energy (American Psychiatric Association, 2013). A total of 90% of depressed humans report feelings of fatigue or loss of energy (Ghanean, Ceniti & Kennedy, 2018). As fatigue is a subjective experience, it is impossible to observe it directly in humans. However, reported fatigue is associated with a decrease in physical activity in humans, although the direction of causality remains unclear (Puetz, 2006).

NHPs likely can experience fatigue, because there is homology in the mechanisms responsible for affect across primates (Bliss-Moreau & Rudebeck, 2021; Lagisz et al., 2020). Accordingly, decreased locomotion is typically used as an indication of fatigue in NHPs (Li et al., 2020; Teng et al., 2021). However, locomotion is also used to assess psychomotor retardation. So, it is not possible to distinguish fatigue from psychomotor retardation. Similarly, fatigue is associated with lower levels of sleep (e.g., Darwent et al., 2015), but this is already another criterion of depression (see “Differences in appetite or weight (S3)”). Thus, although NHPs may experience fatigue, no measures to independently assess fatigue currently exist.

Feelings of worthlessness or guilt (S7)

The DSM-5 names feelings of worthlessness or excessive or inappropriate guilt as a symptom of depression, disclosed by the patient through self-report (American Psychiatric Association, 2013). Feelings of guilt are more prominent in humans with depression than controls (Luck & Luck-Sikorski, 2021). Furthermore, humans with depression have feelings of exaggerated responsibility for uncontrollable events (Kim, Thibodeau & Jorgensen, 2011). In human children and adults, feelings of guilt have been linked to observations of increased reparative prosocial behaviours, such as apologising or stating concern (Bybee, Merisca & Velasco, 1998; Donohue & Tully, 2019). Subsequently, it has been suggested that prosocial approach behaviours can be used as an indicator of feelings of guilt in non-human species (Malhotra, 2019). However, social behaviours can also be diminished as a result of anhedonia (see “Anhedonia (S2)”). Furthermore, the DSM-5 states that the guilt must be excessive or inappropriate (American Psychiatric Association, 2013), so just assessing guilt would not be sufficient. The next step requires to examine when guilt is excessive, which is currently impossible in non-verbal animals. Altogether, feelings of worthlessness/guilt currently cannot be recognised in NHPs.

Diminished ability to think or concentrate, indecisiveness (S8)

The DSM-5 states that an impaired ability to think, concentrate, or make decisions (referred to as cognitive deficits) is often present in depression (American Psychiatric Association, 2013). This is often assessed by self-report, but distraction or impaired “short-term or working” memory can be observed (American Psychiatric Association, 2013; de Almeida et al., 2012; Dillon & Pizzagalli, 2018; Douglas et al., 2009; Marazziti et al., 2010). The DSM-5 uses the term “think” in this criterion. However, it has not been decisively established whether non-human animals are able to think, i.e., reflect on their behaviour, situation or emotions (e.g., Phillips et al., 2021; but see Kano & Call, 2021). Because of this, we will use the term “processing information” instead of “thinking” for NHPs.

A measure of short-term and working memory is the delayed matching-to-sample task (DMTS), in which depressed humans make more errors (Douglas et al., 2009; Porter et al., 2003; Shah et al., 1999), and which can also be used in NHPs (e.g., Chen et al., 2023; Truppa et al., 2014). In a DMTS, a stimulus is presented to the subject. After a short interval, the initial stimulus is presented along with a novel stimulus. The subject is instructed (for humans) or rewarded (for NHPs) to select the initial stimulus (Chudasama, 2010). Although the DMTS has never been used to study depression in NHPs, the DMTS could be a valid indicator of cognitive deficits. A downside of the DMTS is the need for training the participating animals, requiring an extensive time and labour investment. Additionally, because the DMTS measures both short-term and working memory, effects on either of those are impossible to distinguish, Thus, although a valid task to measure cognitive deficits, the use of the DMTS has some drawbacks.

An alternative short-term memory task that requires no training and only comprises two trials is a preferential looking task (see “Depressed mood (S1)”) with two neutral stimuli (Ennaceur & de Souza Silva, 2018), sometimes referred to as a recognition memory task (Capitanio, 2021). In the first trial, subjects are familiarised with a stimulus. In the second trial, the familiar stimulus is presented next to a novel stimulus. A healthy individual spends more time looking at the novel stimuli, whereas NHPs with adverse experiences do not spend more time looking at the novel stimuli (Capitanio, 2021), similar to a paradigm to study recognition memory in human infants (Fantz, 1964; Rose, Feldman & Jankowski, 2004). In humans, results are conflicting: some researchers find that depression negatively impacts recognition memory of previously shown words (Deijen, Orlebeke & Rijsdijk, 1993; Watts, 2014), whereas others do not find a relationship (Brand, Jolles & Gispen-de Wied, 1992; Dunbar & Lishman, 1984). In sum, due to conflicting results in humans, it is unclear whether recognition memory tasks in NHPs validly represent cognitive deficits associated with depression.

In summary, cognitive deficits are measurable in NHPs, but not by mere observation. The delayed matching-to-sample task is a valid measure of impaired cognitive function. The preferential looking task with neutral stimuli might be more feasible, but validity is currently lacking.

Suicidal thoughts, ideation, plans or attempts (S9)

Humans with depression may think about death more often, may be wishing not to be alive, or may even plan and attempt suicide (American Psychiatric Association, 2013). Suicidal ideation is typically elicited through interviewing patients and is present in 11–63% of the depressed population (Vuorilehto et al., 2014). For suicidal behaviour to occur, an understanding of death is required. NHPs have a basic understanding of the concept of death of others, but it remains unclear whether NHPs have a causal understanding of events leading to death (Gonçalves & Carvalho, 2019). This means we cannot assume that NHPs have “thoughts” about death or suicide, and that we should not attribute self-harming or self-endangering behaviours (Walker et al., 2012) to suicidal ideations in NHPs. In sum, there is no measure to assess whether NHPs are capable of having suicidal ideations, so suicidal ideation cannot be used as a criterion of depression in NHPs.

Conclusion: proposed diagnostic criteria of NHP depression

In short, we consider six DSM-5 signs of depression observable in NHPs: depressed mood, anhedonia, appetite/weight changes, diurnal rhythm changes, psychomotor changes, and cognitive deficits. Three DSM-5 signs are not observable: fatigue/loss of energy, feelings of worthlessness/guilt, and suicidality. Therefore, directly applying the human diagnostic criteria for diagnosing depression (at least one core symptom alongside four other symptoms; American Psychiatric Association, 2013) to non-humans, would likely result in a large number of false negative diagnoses. To prevent this, we propose less “strict” diagnostic criteria for non-human depression, compared to the human criteria. We propose the following diagnostic criteria for depression in non-human primates:

A non-human primate is considered depressed when it shows at least one core sign of depression (depressed mood or anhedonia), alongside at least three other signs (appetite/weight changes, diurnal rhythm changes, psychomotor changes, cognitive deficits) during the same two-week period.

In the next part of this review, we will apply the proposed diagnostic criteria to cases of potentially depressed NHPs.

Experimental induction of signs of depression

In the 1960s and 1970s, induction of depression in NHPs became a topic of interest (Harlow & Suomi, 1974). Signs of depression were induced in young rhesus macaques by separating them from their mothers and through use of social isolation and social deprivation. In addition, manipulation of social hierarchies (Shively et al., 1997) and the photoperiod (Qin et al., 2015) led to signs of depression. In this section, we explore whether animals subjected to these experiments show depression following our criteria.

Naturally occurring signs of depression

Besides experimental contexts, signs of depression can also occur naturally. Here, we discuss signs of depression that are associated with maternal loss, natural social separation, and social status loss.

Recognising signs of depression in NHPs

Four or five signs of depression were identified across the reviewed studies. Six out of nine DSM-5 symptoms of depression have corresponding observable signs in NHPs: depressed mood, anhedonia, differences in weight, differences in diurnal rhythm, changes in psychomotor speed, and cognitive deficits (see Table 1). Based on the observable signs, diagnostic criteria for NHP depression were formulated, requiring at least one of the core signs, depressed mood or anhedonia, to be present, alongside at least three other signs of depression during the same 2-week period. Based on the proposed criteria, we found experimental evidence for anhedonia, eating/weight changes, changes in sleep, and psychomotor speed changes. We found anecdotal evidence for cognitive deficits, and no evidence for the presence of depressed mood (see Table 2). In experimental contexts, anhedonia, eating/weight changes, changes in sleep, and psychomotor speed changes were observed. For naturally-occurring depression, only studies with lower levels of evidence were available. We found possible evidence for anhedonia, correlational evidence for weight changes and anecdotal evidence of psychomotor changes and cognitive deficits. Signs often persisted for longer than 2 weeks in both experimental and non-experimental contexts (e.g., Goodall, 1986; Qin et al., 2015; Shively et al., 1997; Spencer-Booth & Hinde, 1971; Suomi & Harlow, 1972; see Table 2). However, only one group of NHPs in the reviewed studies (Boccia et al., 1997) met our diagnostic criteria for depression.

Why were diagnoses of NHP depression so rare?

The question then arises why so few diagnoses of NHP depression were made in this review. First, we reiterate that the lack of diagnoses is not evidence of absence of depression in NHPs; we argue that the lack of diagnoses can mainly be explained by the lack of research into the co-occurrence of all depressive symptoms. Although six signs of depression can be observed in NHPs (as outlined above), no empirical study has thus far examined all six signs simultaneously. Previous research typically focussed only on easily observable signs, such as hunching behaviour, weight differences or changes in locomotion. In contrast, measuring depressed mood or cognitive deficits is more difficult, as this requires setting up a task. Indeed, no study explored these symptoms. We recommend future researchers to examine the co-occurrence of all measurable signs of depression during a period of 2 weeks or more. Considering that using suboptimal methods, some NHPs could still be diagnosed with depression (Boccia et al., 1997), we expect that more research into co-occurrence of the observable signs will result in more diagnoses of depression.

Another limitation contributing to the lack of diagnoses is that signs of depression are often measured and analysed on group level instead of on individual level. In most experimental studies, the aim is to find out whether certain circumstances can induce depression, which is analysed on the group level (Harlow & Suomi, 1974; Hennessy, Chun & Capitanio, 2017; Qin et al., 2015; Shively et al., 1997). This is in contrast with our interest: the individual behaviour and subsequent diagnosis.

Limitations and strengths

To formulate our diagnostic criteria of NHP depression, we translated human DSM-5 criteria of depression to observable behaviours in NHPs. We operationalised the signs of depression in concrete and observable behaviours across different species. Compared to the human diagnostic criteria, our NHP criteria may be too loose: our diagnostic criteria require the presence of four out of six observable signs of depression, compared to five out of nine for humans. However, we consider the less strict diagnostic criteria justified to prevent false negative diagnoses, since three out of nine symptoms of human depression have no equivalent observable behaviours in NHPs.

In addition, although this review spans a range of primate taxa, 16 out of the 21 reviewed studies are on macaques. So, this sample does not equally represent the entire primate order. Similarly, the behavioural measures of signs of depression were mostly based on studies in macaques and apes. Therefore, some caution is advised when applying methods to taxa that they were not directly developed for.

In experimental settings, within-individual or between-group comparisons can be made to study the differences in signs of depression across conditions. However, in the reviewed cases of naturally occurring signs of depression, no pre-depression data or control groups are available for comparison. Thus, these conclusions rely on anecdotal reports or correlational evidence rather than experimental evidence. The lower level of evidence in the studies in non-experimental contexts excludes causal inferences and limits the reliability of assessments.

We aimed to diagnose depression using minimally labour intensive, minimally intrusive, and most easily accessible methods. However, for some signs of depression we could not reach these goals. For example, assessing depressed mood and cognitive deficits requires tasks.

Furthermore, although we identified several potential risk factors of signs of depression in NHPs, our review does not provide a comprehensive assessment of the prevalence or severity of each risk factor. We only included studies where signs of depression were observed, which yields an unrepresentative sample for comparing risk factors.

A strength of our review is the addition of new insights into non-human depression, adding onto other work in the field of NHP depression research. Compared to previous depression research on chimpanzees (Ferdowsian et al., 2011), we argue that our criteria are more concrete and only directly observable behaviours are included. Compared to a recent review on animal models of human mood (Bliss-Moreau & Rudebeck, 2021), our review focused on criteria for diagnosing DSM-5-derived signs of depression in NHPs through behavioural observation, rather than focussing on neurobiology and affective states. Another review identified similarities in animal models of depression and common captive animal housing (Lecorps, Weary & von Keyserlingk, 2021). We added to this the recognition of specific signs of depression associated with other risk factors, in both wild and captive conditions. A different review focused on depression-like behaviours in NHPs (similar to our article), along with anxiety-like behaviours (Ausderau et al., 2023). The latter two reviews (Ausderau et al., 2023; Lecorps, Weary & von Keyserlingk, 2021), however, were not focused on diagnosing NHP depression, but more on broadly exploring which NHP behaviours are associated with depression, and on reflecting on the ethics of researching depression-like and anxiety-like behaviours in NHPs. Our review added to this by providing clear diagnostic criteria of NHP depression and in providing a translation to observable behaviours in NHPs. Another recent review (MacLellan et al., 2021) showed parallels with our method, but there are differences in diagnostic criteria. For example, we added anticipatory behaviour as a measure of anhedonia and use different methods to assess cognitive deficits. Moreover, whereas this review concerned depression in poor housing conditions (MacLellan et al., 2021), we included a broader set of risk factors of depression in NHPs. In sum, our work added onto previous work by providing a new approach to NHP depression.

Future research

First of all, more research into observing co-occurrence of signs of depression in individual NHPs is needed. In the reviewed literature, typically only a few signs were measured on group level. To assess depression in individual NHPs, as many signs as possible must be measured on an individual level. This may come with difficulties, as this requires ongoing baseline data collection to compare within-individuals when depression is suspected. An alternative may be assessing baseline levels of depressive parameters in healthy individuals, which also allows recognition of species-specific differences in the expression of signs of depression. In addition, age differences in depression as described in humans, such as irritability as a sign of depression in young individuals (Sherwood et al., 2021), might also be discovered. Based on these data, deviations from the species baseline can be identified. Nevertheless, truly objective assessment of depression (Rosati et al., 2012) requires experimental studies, so this cannot be attained in principle for naturally occurring depression. If research into the co-occurrence of signs is conducted, we expect that more NHPs will be diagnosed with depression according to our criteria.

Furthermore, we encourage further development of methods measuring signs of depression. First, we recommend research to potentially identify easily observable behaviours that can serve as a proxy for depression or for signs that currently require tasks to measure. These methods should take species-specific social behaviour, signals and reactions into account. Second, we encourage research into observing the three signs of depression that are currently not (independently) observable: fatigue, feelings of worthlessness/guilt, and suicidality.

In addition, researching the effect of antidepressants on the proposed indicators of in NHPs can help improve the validity of these indicators. Although there is recent criticism on the efficacy rates of antidepressants (Moncrieff, 2018) and antidepressants can also be used to treat other conditions (Bandelow et al., 2015), exploring the effect of antidepressants may provide valuable insights. Additionally, it is important that depression assessment is done by experts who are familiar with the species and behavioural observations, because possible interventions, like prescribing antidepressants, should only be carried out when there is a strong basis to do so.

The administration of antidepressants is one intervention that can reverse signs of depression in experimentally induced depression (e.g., Harlow & Suomi, 1974; Qin et al., 2015) and even in naturally occurring depression (Chu, 2019). Alternatively, some types of enrichment can reduce signs of depression (MacLellan et al., 2021). Other interventions may be to address the risk factors, if possible. If an individual expresses (signs of) depression while being housed without access to a social group (Hennessy, Chun & Capitanio, 2017; Jackson et al., 2023; Li et al., 2013), the opportunity for social interactions could be increased, for example in social group housing. This is in line with legislation in the European Union, where social housing of primates is obligatory (EU Directive 2010/63). Evidently, not all risk factors can be prevented, such as status loss (Kummer, 1995; Setchell & Dixson, 2001) or maternal death (Goodall, 1986). Thus, for some risk factors, a diagnosis of depression could aid in assessing the severity of the effect of the risk factor and allow for interventions.

Due to our inclusion criteria, our review could not provide an overview of the prevalence or severity of the risk factors. Therefore, we recommend future research specifically into risk factors of (signs of) depression in NHPs. Furthermore, although the scope of our review was limited to signs of depression attributable to one event, our diagnostic criteria are also applicable to animals subjected to an accumulation of stressors. Examples include consistent maltreatment by humans (Lopresti-Goodman, Kameka & Dube, 2012), insufficient housing conditions (Lecorps, Weary & von Keyserlingk, 2021), repeated loss (Rasmussen & Reite, 1981), and chronic stress (Teng et al., 2021; Wang et al., 2013; Zhang et al., 2016). Evidently, and beyond the scope of our review, there are non-behavioural parameters that can help identify depression. Biomarkers such as the levels of pro-inflammatory cytokins (MacLellan et al., 2021) or heart rate variability (Jarczok et al., 2018) are associated with depressive symptoms in humans and signs of depression in NHPs. For assessing sleep changes, electroencephalograms (EEGs) are widely used in NHPs (Li et al., 2022; Reinhardt, 2020). Ideally, behavioural research into NHP depression should complement and be complemented by research into non-behavioural parameters associated with depression.