SEARCH

SEARCH BY CITATION

Keywords:

  • Addiction;
  • affective disorders;
  • animal communication;
  • autism;
  • bioacoustic communication;
  • empathy;
  • mood;
  • schizophrenia;
  • ultrasonic vocalizations

Abstract

  1. Top of page
  2. Abstract
  3. Prosody
  4. Genetics
  5. Future directions
  6. Summary
  7. References
  8. Acknowledgments

Mental illness can include impaired abilities to express emotions or respond to the emotions of others. Speech provides a mechanism for expressing emotions, by both what words are spoken and by the melody or intonation of speech (prosody). Through the perception of variations in prosody, an individual can detect changes in another's emotional state. Prosodic features of mouse ultrasonic vocalizations (USVs), indicated by changes in frequency and amplitude, also convey information. Dams retrieve pups that emit separation calls, females approach males emitting solicitous calls, and mice can become fearful of a cue associated with the vocalizations of a distressed conspecific. Because acoustic features of mouse USVs respond to drugs and genetic manipulations that influence reward circuits, USV analysis can be employed to examine how genes influence social motivation, affect regulation, and communication. The purpose of this review is to discuss how genetic and developmental factors influence aspects of the mouse vocal repertoire and how mice respond to the vocalizations of their conspecifics. To generate falsifiable hypotheses about the emotional content of particular calls, this review addresses USV analysis within the framework of affective neuroscience (e.g. measures of motivated behavior such as conditioned place preference tests, brain activity and systemic physiology). Suggested future studies include employment of an expanded array of physiological and statistical approaches to identify the salient acoustic features of mouse vocalizations. We are particularly interested in rearing environments that incorporate sufficient spatial and temporal complexity to familiarize developing mice with a broader array of affective states.

Many of life's enriching experiences occur within a social context, with the ability to share enjoyment, to grasp how another individual feels, or to see the world from the perspective of another. In a variety of developmental disabilities and neurological diseases, these capacities are diminished or lost (Abdi & Sharma 2004, Freeman et al. 2009, Hagerman et al. 1986, Lord et al. 2000, O’Keeffe et al. 2007, Orbelo et al. 2005, Ruby & Decety 2004, Testa et al. 2001). To elucidate how these deficits emerge in development, adulthood or aging, emotional experiences cannot be directly measured. Rather, changes in behaviors are used as a metric to infer changes in emotion. For example, the level of fear experienced by a person cannot be measured but the quiver of a fearful voice can be detected. Scientists can also measure behavioral indications of empathy; how one individual responds to the emotional expressions of another individual. For instance, in response to hearing the quiver in a voice, an individual might feel compassion, fearful, vindicated, frustrated or angry. In turn, these emotions can engender various behavioral responses, such as expression of consoling words, flight from the scene, pursed lips, or a confrontation with the dangerous stimulus. The purpose of this review is to explore how mouse vocalizations convey emotional information, how mice respond to these vocalizations, and how we can infer affective states from changes in the mouse vocal repertoire.

Prosody

  1. Top of page
  2. Abstract
  3. Prosody
  4. Genetics
  5. Future directions
  6. Summary
  7. References
  8. Acknowledgments

Vocalizations provide highly salient cues that can indicate changes in emotion. Emotional information can be provided by what words are spoken (from the meanings of the words themselves). ‘That feels great!’‘You’re making me frustrated’. ‘That dude gives me the creeps'. We know the meaning of these statements and we often express an emotional response to them. But emotions are also expressed by how words are spoken. This quality of speech, which is called ‘prosody’ and is often referred to as intonation, is central to communication. Prosody is measured by variations in the pitch (notes), energy (loudness), and the duration of spoken words and the timing between them.

Prosody can be classified into three functional groups. Grammatical prosody entails changes in emphasis on particular syllables, or changes in pitch, within a sentence that can help us distinguish meanings. For example, applications of stress on one of two syllables in a word can help us to differentiate a dry, treeless ecosystem (des'-ert) from the pleasant plate of food at the end of a meal (des-sert'). An upward trend in pitch through a sentence enables us to discern a question from an assertion. The soup bowl is in the kitchen. The soup bowl is in the kitchen? Pragmatic prosody is used to help us put emphasis on an important idea, sentence or word in a sentence, and thus shows our focus and intentions. For example, enhanced duration and loudness of a particular word can indicate its importance within a sentence. ‘I don't want the brown coat, I want the DOWN coat'. A different function of pragmatic prosody is that it signals the context, roles, and relationships between speakers. People speak to infants in ‘motherese’ or ‘infant-directed talk’ in a voice that they do not use with older children (Fernald 1989). People speak with one tone to their employer and a different tone with their intimate partner. Affective prosody is responsive to internal emotional experiences, such as sadness, anger, fatigue, annoyance, concern, contentment, relief, excitement or victory. For instance, high levels of frequency modulation, in which there are broad variations in the level of pitch, from syllable to syllable, can be associated with excitement. A minimal level of frequency modulation, or monotone, can indicate boredom or fatigue. Prosody is also responsive to social context. Adults talk with old friends and new acquaintances with different vocal patterns. The variability of speech within the context of changes in emotion or social context is referred to as emotional or affective prosody.

Affective prosody and mental disorders

Autistic individuals have been shown to assign stress to the wrong syllables of a word (Baltaxe 1984, Baltaxe & Guthrie 1987, Shriberg et al. 2001) and children with high-functioning autism can have difficulties modulating the pitch and controlling the volume of their speech (Shriberg et al. 2001). Some aspects of autistic speech incorporate all three features of the diagnosis: repetitive behaviors, deficits in the ability to express emotions and deficits in communication. For instance, some children with autism repeat the use of certain sounds, syllables or words, more than typically developing children (Shriberg et al. 2001), or fail to use appropriate patterns of intonation to communicate. Prosody can be monotonic, minimally pitched or energy modulated, or it can be amplified in pitch range or even singsong-like (Amorosa 1992), masking dynamics in emotional status. Autistic language may not be appropriately attuned to context, or ‘machine-like’ (Fine et al. 1991). Overall, speech impairments in autism are generally more prevalent in the expression of pragmatic or affective prosody than grammatical prosody (Shriberg et al. 2001). Individuals with an autism diagnosis can have impairments in the ability to understand the spoken language and gestures of others, even as infants (Golan et al. 2007, Kuhl et al. 2005, Rutherford et al. 2002). The ability to respond to the emotional expressions of others can be deficient among people with autism (Silani et al. 2008, Smith 2009). In this regard, autistic children also show deficits in their abilities to use prosodic cues to disambiguate sentence meaning (Diehl et al. 2008). These deficits can extend to inabilities to interpret intonations that denote liking or disliking (McCann et al. 2007).

Schizophrenia, mania and depression can also be significantly impaired in the ability to comprehend affective prosody, while schizophrenia can additionally feature substantial deficits in the ability to express affective prosody (Murphy & Cutting 1990). Schizophrenia can feature deficits in expressive and receptive prosody (Bach et al. 2009, Leentjens et al. 1998, Leitman et al. 2006), especially in paranoid (Bach et al. 2009) and flat affect (Alpert et al. 2000) schizophrenia. For instance, individuals with flat affect schizophrenia employ normal word patterns to express emotions (frequency of words describing pleasurable or distressful experiences), but they communicate these experiences with less prosodic inflection (Alpert et al. 2000). Like autism disorders, there are many forms of schizophrenia which do not all show deficits in the expressive prosody (Cohen & Docherty 2005, Wan et al. 2008). Receptivity to emotional expressions of others can also be impaired in schizophrenia (Derntl et al. 2009, Fujiwara et al. 2008, Haker & Rossler 2009), particularly in males who are responding to emotions that have negative valence (Bozikas et al. 2006, Pijnenborg et al. 2007, Scholten et al. 2008). In addition to schizophrenia, responsiveness to affective prosody can be impaired in depression (Emerson et al. 1999, Uekermann et al. 2008), multiple sclerosis (Beatty et al. 2003), Parkinson's disease (Benke et al. 1998), alcoholism (Monnot et al. 2001, 2002, Uekermann & Daum 2008, Uekermann et al. 2005) and following exposure to MDMA or ‘ecstasy’ (Yip & Lee 2006).

Mouse vocalizations

To model deficits in expressive and receptive prosody, there must be evidence that features of mouse vocalizations typically associate with specific affective states, such as distress, or reward, or anticipation of reward and whether these features can be sensitive to differences in specific alleles or genetic background. Mice have the capability to vocalize across a broad range of frequencies that extend from as low as the human-audible range (when we hear squeaks) to well into the ultrasound range, above the limit of human hearing [20 000 cycles per second (hertz) = 20 kHz]. Audible squeaks are produced by laboratory mice in stressful and painful situations (Whitney & Nyby 1983), such as during handling and restraint (Whitney 1969), grid-shock test (Winter 1965) or during aggressive encounters (Gourbal et al. 2004, Houseknecht 1968). In reproductive contexts, human-audible squeaks are produced by females when a sexually motivated male is interacting with a non-receptive female (Sales 1972).

Over the last 40 years, mouse vocalizations have been studied mostly within two social contexts: adult male vocalizations emitted in response to the cues of estrous females and vocalizations emitted when pups are separated from the nest. Initial analyses of these calls were limited to rate of call production within a particular bandwidth of frequencies. More recently, analyses include classification of discrete vocalizations based upon their differences in frequency (pitch) and energy (loudness).

Adult male–female interactions

Adult vocalizations have been primarily studied within reproductive contexts. Adult males emit ultrasonic vocalizations (USVs) when exposed to a female partner or to her odor cues (Maggio et al. 1983, Nyby 2001, Whitney & Nyby 1979). During courtship, the male vocalizes and approaches the female (Maggio et al. 1983, Whitney & Nyby 1979). The relationship between male USV production and female behavior suggests that USVs predict mating opportunities rather than aggression. Female responses to these USVs include attenuated aggressive behaviors, approach and more frequent expression of mating postures (Sewell 1969). In a two-compartment choice test, females approach tethered males that are vocalizing vs. males that are surgically rendered incapable of vocalizing. If both males are rendered unable to vocalize, but one of the males is placed next to an ultrasound generator producing 70 kHZ ultrasounds, females prefer to reside in the compartment containing the male with the sound generator (Pomerantz et al. 1983).

During these male–female social interactions, males emit two types of vocalizations (Maggio et al. 1983, Nyby & Whitney 1978, Wang et al. 2008, Whitney & Nyby 1979). One is a highly modulated pure tone with a mean frequency of around 70 kHz and an intensity of 40 dB (Sales & Pye 1974). The second is a pure tone around 40 kHz with an intensity of more than 100 dB. The male emits the 70 kHz calls almost continuously prior to the first mount, while he actively sniffs and investigates the female. After the first mount, the number of calls declines, although during the latter stages of copulatory behavior, the male intermixes both 70 and 40 kHz calls immediately before and between mounting bouts (Nyby 1983, Sales & Pye 1974, White et al. 1998).

The development of digital sound spectrographic analysis of USVs has allowed researchers to expand analyses of mouse vocalizations to include dynamic changes in mean frequency (pitch). Individual calls can then be classified according to changes in dominant frequency, as a function of time. This analytical approach was used to assess USVs of adult males that were elicited by odors of receptive adult females (Holy & Guo 2005). Recordings were collected and calls were classified. For each call, pitch changes were compared for sudden punctuations in pitch that occur within a syllable. This analysis showed distinct clusters of pitch changes, such as upward and downward jumps between comparatively low frequencies (35–50 kHz) and similar transitions from between 70–90 kHz and 55–70 kHz (see Fig. 2e,f). There were also calls that lacked any abrupt changes in pitch. Often these calls were emitted in repeated temporal sequences (Holy & Guo 2005) and changed reproducibly with the succession of approach, mounting, and release behaviors that constitute the mating encounter. Before mounting, males usually called in either flat or continuous frequency-modulated USVs. However, when males mounted females, males shifted to frequency-modulated broken ‘step-like’ USVs with 70-kHz and 40/80 harmonic frequencies until the end of mounting (Wang et al. 2008).

image

Figure 2. Spectrograms of representative call types emitted by juvenile male C57BL/6J mice (P30) mice reunited in a home cage environment after 24 h of social isolation. Spectrograms include examples of an upward-modulated call (a), downward-modulated call (b), chevron (c), complex call (d) and punctuated call (e). The abrupt change in frequency in figure e is a pitch jump (p), indicated by the downward pointing arrow (f). Time (in seconds) is indicated by the X-axis, frequency in kHZ is indicated by the Y-axis, and relative intensity or loudness is indicated by color (see key).

Download figure to PowerPoint

Could these male vocalizations be relevant to affective prosody? Inferences that males derive a feeling of reward from social interactions with estrous females are warranted, based upon studies that use conditioned place preference (CPP) testing (and see below for discussion of CPP). Specifically, male mice prefer environments that have been associated with either mating (Kudwa et al. 2005, Popik et al. 2003) or even access to (Pankevich et al. 2006, Pierman et al. 2006) estrous female mice. The changing repertoire of male USVs through the mating process could reflect a shift from anticipation to consumption of sexual reward. In other words, these studies suggest that male mice may emit different peak frequencies that correlate with incentive salience vs. hedonic pleasure (see Berridge 2007).

Adult same-sex interactions

During male–male agonistic encounters, two types of calls can be expressed. V-shaped USVs encompass a broad frequency range and are emitted during naso-nasal contacts and body sniffing behaviors that precede fighting. A second set of calls consists of harmonics, or overtones, which are recorded during fighting behavior (Gourbal et al. 2004). These results are in agreement with the previous studies (Scott 1966) including observations that the number of audible vocalizations emitted during a male–male encounter correlates with time fighting and number of attacks (Brain et al. 1980, Morgret & Dengerink 1972). These audible vocalizations emitted during solicitation and mating, as there were similar to the ultrasonic vocalizations, are differences in USV structure that are associated with the anticipation vs. participation in social interactions. Taken together, the differences in frequency modulation associated with male anticipation and consumption of sexual vs. aggressive encounter indicate that these call patterns respond to differences in the affective nature of the social encounter.

There are fewer studies of vocalizations among female mice. Females express audible low-frequency vocalizations during copulation, particularly when they are not in estrous, consistent with the observation that they tend to avoid sexual encounters with males unless they are reproductively receptive. By contrast, female–female encounters engender USV production, especially when they were engaged in olfactory investigation of the other (Sewell 1970). Female mice emit a large number of USVs, typically emitted at 70 kHz, during the first minutes of social interaction (Gourbal et al. 2004, Moles et al. 2007). The rate of USV production among females appears to be responsive to levels of social motivation, experimentally imposed by food deprivation, estrous cycle, pregnancy and aging, such that USV rates positively correlate with levels of social investigation (Moles et al. 2007). Interestingly, just as estrous females produce fewer calls during same-sex interactions, brief exposure to a male also inhibits USV production (Maggio & Whitney 1985).

USVs of infant mice and the maternal response

Infant mice of postnatal day (PD) 1–4 are born blind and deaf (Fox 1965), have limited motor abilities, lack fur and subcutaneous fat and cool rapidly when displaced from the nest. When removed from the nest, they emit separation calls that elicit female retrieval. Under conditions of stress, cold and hunger, pups emit calls that engender maternal nest building, licking behaviors, and crouching behaviors that allow for access to teats (Noirot 1969). Rates of separation calls change with age and peak between PD 3 and PD 7, depending upon background strain (Branchi et al. 1998, Elwood & Keeling 1982, Hennessy et al. 1980, Scattoni et al. 2008b). By two weeks of age, juvenile mice no longer emit separation calls (Elwood & Keeling 1982, Noirot 1966).

As early as 1970, Sewell showed that lactating female wood mice (Apodemus sylvaticus) more frequently enter a compartment containing a loudspeaker emitting the recorded USVs vs. background noise or an artificial stimulus (Sewell 1970). Early studies also showed that lactating Mus musculus females preferentially approach ultrasounds within the frequency range of the natural calls (ranging from 40 to 80 kHz) vs. other ultrasonic sounds (Ehret & Haack 1981). More recently, a strain-dependent difference in the dynamic relationship between maternal responsiveness and pup calling rate was found. Using a three-compartment test structure, in which the mother could only reach her pups by crossing the central chamber which contained olfactory cues from a potentially infanticidal male, C57BL/6 mothers expressed higher maternal responsiveness vs. BALB/c females, and their pups emitted fewer calls than BALB/c pups (D’Amato et al. 2005). These data suggest that maternal responsiveness to pup calls, in turn, affects the rate that pups emit USVs.

Several studies show that USV production in response to separation and isolation can be a physiological response to a thermal challenge, e.g. a reflexive abdominal compression reaction in response to the cold that helps return venous blood to the heart (Blumberg & Alberts 1990, Blumberg & Sokoloff 2001). Additionally, pups vocalize when placed in a warm location (Branchi et al. 2004, Shair et al. 2003, Wöhr et al. 2008). In these cases, studies that employ knockout mice with alleles relevant to social bonding and separation distress suggest that separation calls are generated in response to affective changes. For instance, mice that lack a functional oxytocin allele, a molecule that plays a role in social bonding (Ferguson et al. 2000, 2002, Kavaliers et al. 2004, Mahler & Berridge 2009, Petrovic et al. 2008), emit fewer separation calls than their wild-type littermates (Winslow et al. 2000). Rates of separation calls are also influenced by the presence or absence of a mu-opioid receptor (Moles et al. 2004), which is a well-defined mediator of reward physiology and separation distress in rodents (Mahler & Berridge 2009, Panksepp et al. 1980).

Spectrographic analysis has showed that neonatal laboratory mice emit different kinds of USVs in response to varied environmental conditions to which the pup is exposed, such as odor from the nest, social isolation, low temperature, tactile stimulation or odor of an unfamiliar conspecific adult male (Branchi et al. 1998). For instance, isolated pups emit a high percentage of calls that contain abrupt breaks in frequency steps but an equal number of gradually modulated frequency calls relative to pups exposed to the odor of an unfamiliar, potentially infanticidal, adult male (Branchi et al. 1998). Variations in USV frequency modulation are also sensitive to the age of the pups, such that calls become stereotyped with maturation but remain acoustically distinct from adult USV patterns (Liu et al. 2003). See Fig. 1 for comparison of mouse vocalizations across development.

image

Figure 1. Spectrograms of call sequences differ among mice of varied ages and social contexts. Pup separation calls were collected from an 8-day-old mouse of the C57BL/6J strain approximately 1 min after removal from the nest and placement in a soundproof chamber at 23°C. Wriggling calls were recorded from pups within the litter of the C57BL/6J strain approximately 1 min after placing the microphone above the nest. Juvenile calls were recorded approximately 30 seconds after two male C57BL/6J mice (PD30) were reunited in a home cage environment after 24 h of social isolation. Calls emitted during female–female interactions were collected approximately 20 seconds after an intruder female was inserted into the cage of a resident female, both mice of the C57BL/6J strain after 3 days of social isolation. Calls emitted during male–female interactions were collected approximately 20 seconds after an adult female of the C57BL/6J strain was inserted into the cage of an adult male mouse of the C57BL/6J strain in his home cage environment. Calls emitted during male–male interactions were collected approximately 20 seconds after an intruder male of the C57BL/6J strain was inserted into the cage of a resident male mouse of the C57BL/6J strain in his home cage environment after 3 days of social isolation.

Download figure to PowerPoint

Mouse pups also produce ‘wriggling’ calls when struggling in the nest, mainly when pushing for the teats during suckling by the mother (Ehret 1975). These calls release at least three types of maternal behavior that include licking of pups, changes of suckling position, and nest building (Ehret & Bernecker 1986). Wriggling calls usually consist of a fundamental frequency near 4 kHz and overtones that can extend to a maximum frequency of about 20 kHz (Ehret & Bernecker 1986, Geissler & Ehret 2002, 2004). These calls elicit a specific neural activity. There is a differential response of C-fos immunoreactivity within the auditory cortex to synthetic wriggling calls played back to nursing females vs. similar synthetic wriggling USVs with a temporally displaced harmonic (Geissler & Ehret 2004). Interestingly, differences in immunoreactivity were more pronounced in the left hemisphere (Geissler & Ehret 2004), consistent with electrophysiological studies (Stiebler et al. 1997) and sharing similarities with human auditory processing (Ehret & Riecke 2002).

Genetics

  1. Top of page
  2. Abstract
  3. Prosody
  4. Genetics
  5. Future directions
  6. Summary
  7. References
  8. Acknowledgments

Call rates and frequency modulation of male USVs are responsive to genetic manipulations of endogenous reward pathways. For instance, targeted deletion of muscarinic receptor M5 results in an 80% decrease of male USVs production during sexual interaction. Male exposure to amphetamine, a direct ligand of D2 receptors, results in increased USV production, also responsive to targeted disruption of muscarinic receptors (Wang et al. 2008). Amphetamine can modulate dopamine release in the nucleus accumbens (Yeomans et al. 2000, 2001), thereby exerting a substantial role in appetitive behaviors (Berridge 2007, Ikemoto & Panksepp 1999, Kelley & Berridge 2002). Similarly, infant mice lacking functional mu-opioid receptors express reduced numbers of USVs during maternal separation (Moles et al. 2004), consistent with the evidence that enkephalin interactions with mu-opioid receptor can mediate social reward and isolation distress (Bertrand et al. 1997, Kalin et al. 1988, Panksepp et al. 1980, 1994, Vanderschuren et al. 1995).

Genetic manipulations that do not directly interact with incentive salience or hedonia can also modify call rate and call repertoire. For example, mice lacking functional components of circuits involved in social recognition, which include oxytocin and vasopressin receptor 1b, emit fewer calls (Scattoni et al. 2008a, Winslow & Insel 2002). Mice that lack functional alleles of genes associated with autistic-like social deficits, such as MECP2 and neuroligin 4 knockout mice, can express a diminished call rate (Jamain et al. 2008, Picker et al. 2006). Classification analysis has also showed that a spontaneous mutation (BTBR) and a targeted allelic replacement (foxp2 R552H) can influence the pup repertoire (Fujita et al. 2008, Scattoni et al. 2008b, Shu et al. 2005). This recent demonstration that ultrasound vocalizations can be modified by manipulations of foxp2, a gene involved in a familial form of speech impairment in humans and required for normal vocal development in songbirds, suggests that rodents may also be an important model for studying the neural and genetic basis of speech learning.

Mouse USVs and affect: theoretical considerations

Changes in subjective affective experience can be a proximal cause for changes in behavior. At a basic level, fear can induce freezing behavior, discomfort can induce withdrawal behavior, and anticipation of reward can induce an approach behavior. To the extent that specific aspects of the USV repertoire are associated with particular affective states, USV studies can provide insight to the genetic mechanisms that underlie affect regulation in normal psychological development and in mental illness.

Among humans, individuals make inferences to the quality of another person's experience based upon verbal communication or based upon observations of an individual's behavior. One means of communicating personal experience is through use of words. Self-report is an essential analytical tool for human research and clinical diagnosis, the basis of tremendous advances in psychological and psychiatric medicine. However, self-report has scientific and philosophical limitations. Bertrand Russell (1912) identified the problem: while words or labels are used to describe universal subjective states, it remains unknown whether these labels identify identical internal experiences. For example, two individuals can use the label ‘red’ to describe an apple but the red apple may appear as red-orange to one person and red-purple to another. It is inherently unknown whether both individuals have identical internal experiences, because of perceptual variations, such as differences in vision or brain function. Differences in sex, age, cultural context and mental illness further contribute to variation in how words label subjective experiences. For instance, post-traumatic stress disorder might impose an atypical sensitivity to stimuli that may be inadequately described with the existing verbal repertoire. Discrete labels used to communicate individual perceptions, no matter how basic or complicated, have limitations in terms of describing internal experience.

To gain greater insight to the affective experience of individuals, new assessment tools have been developed that do not depend upon verbal report. Directionality of eye-gaze (Klin et al. 2002, Wolf et al. 2008) and patterns of brain activity provide insights to how autistic individuals attend to social interactions (Baron-Cohen et al. 1999). Assessments of heart rate variability provide insight to how children experience distress of others (Eisenberg et al. 1988). To gain insight to the etiology of developmental disorders, there are ongoing efforts to identify the psychological experiences that are experiences during mental illness. For example, there is ongoing controversy about whether some forms of autism can result from social anxiety (Kuusikko et al. 2008), deficits in temporal cognition (Boucher et al. 2007) or central coherence (Briskman et al. 2001, Happe et al. 2001). Just as efforts to infer psychological experience can be useful for elucidating mechanisms underlying mental illness, the same issues are important for mouse research. If a knockout mouse expresses low levels of social approach, how do we know whether the targeted allele models autism, vs. depression, vs. shyness?

Such questions are rather new to behavioral neuroscience. Considerable precedence in the history of philosophy, extending back to the work of Rene Descartes (1637), maintains that animals are ‘automata’ or machine-like, in contrast to ‘sentient’ or soulful humans. With the advent of cognitive and affective neuroscience, these Cartesian distinctions are no longer relevant, but it is important to be careful about inferences to rodent affective experience (Panksepp 1998, Schneirla 1959). Such inferences require consideration of the theoretical framework of affective neuroscience, including operant, fear and preference conditioning paradigms, as well as studies of functional anatomy and brain/systems physiology (Panksepp 1998). Such an approach provides working hypotheses for comparative studies of affective prosody that can be subject to falsification.

Operant and place preference conditioning are used to establish anticipation and consumption of rewards (Bardo & Bevins 2000, Kelley et al. 2002, Moles et al. 2004), and aversive conditions can be identified by conditioned fear learning paradigms (Falls et al. 1997, LeDoux et al. 1988, Paylor et al. 1994). Emotional responses to stimuli can also be inferred from studies of functional anatomy, particularly the highly conserved limbic structures in the brain (MacLean 1990) and the underlying physiological systems that mediate affect regulation, including dopamine, serotonin, corticosterone and endogenous opiates (Panksepp 1998). Drugs of abuse, such as opiates and amphetamine, are also useful, because they can influence natural reward systems (Abarca et al. 2002, Bardo & Bevins 2000, Kelley & Berridge 2002, Reith & Selmeci 1992, van Ree et al. 1999). Pharmaceuticals that modulate affective states in humans can be employed that influence analogous behaviors in mouse models (Holsboer 2001, Hunsberger et al. 2007, Kato 2006, Pezet & Malcangio 2004, Rupniak et al. 2001).

In summary, assessments of prosody in mice along with measures of neural and behavioral correlates of motivated behavior, allow for greater reliability in inferences to subjective experience. Such assessments are important for translational applications of mouse research to mental illness. While human vocal patterns are vastly more complex with regard to frequency modulation, timing, overtones, and range of sound produced, and may provide more nuanced emotional content, and studies of human language can assess the temporal associations between prosodic content and referential aspects of language, basic elements of prosodic content in mice are measurable and have been associated with rewarding and aversive conditions. Thus, studies of prosody in humans and mice may yield apples-to-apples comparisons useful for subjective inference.

Mouse USVs and affect: practical considerations

Behavioral associations with USV production

Juvenile mice emit USVs when meeting a littermate after 24 h of social isolation. When USVs of juvenile C57BL/6J (B6) and BALBc/J (BALB) mice are compared during reunion, there are clear differences in the kinds of calls that they emit. For example, there are more downward-modulated calls among reunited B6 juveniles and more upward-modulated and chevron calls among BALB juveniles (see Fig. 2 for example spectrograms).

Differences in USV production between genetic variants of mice could be explained as products of discrete genetic contributions to the anatomies of the larynx or labia. Alternatively, there may be particular calls that are associated with state, such as the motivation to solicit social interaction. To assess the likelihood of these opposing hypotheses, we compared the usage of various call types by different individuals within a strain that express differences in social approach behaviors. Individual B6 and BALB juveniles express more USVs when they are engaged in more vigorous social approach behaviors (Panksepp et al. 2007). During more vigorous social approach behaviors, B6 mice disproportionally express calls that are of a higher pitch and more downward-modulated. These differences in B6 vocal repertoire suggest the possibility that different call types correspond to different degrees of social motivation or function to engender discrete levels of motivation in others. Also important, selective B6 use of particular call types during higher levels of social interaction was specific to B6 mice. BALB mice that engaged in vigorous social interactions expressed all call types at a higher rate. This finding indicates that the degree of repertoire flexibility within a strain is also dependent upon genetic background.

Rodents prefer social encounter to isolation, consistent with classical theories of motivated behavior (Bardo & Bevins 2000, Berridge & Robinson 2003, Tzschentke 1998) that underscore a role for reward and punishment in behavioral approach and withdrawal (Glickman & Schiff 1967, Schneirla 1959, Young 1959). Mouse responses to conditioned place preference (CPP) tests show that they can find social encounters rewarding and social isolation aversive. In a standard CPP experiment, mice are placed in one of two distinct environments. For instance, the beddings of the two environments may be different (one is aspen and the other is paper), the walls may have horizontal vs. vertical lines, or the floors may have circular holes in them or exist as a parallel grid. During the conditioning phases of these experiments, the test mouse is housed successively in either one or the other environment. The mouse has access to a putative reward (e.g. peanut butter, chocolate, cocaine, morphine) in one environment but not in the alternate environment. If the mouse finds the test stimulus rewarding, then through repeated association, the environment associated with access to the reward eventually elicits approach behavior. At the end of the conditioning phase, the mouse is placed in a testing structure that does not contain the reward but contains both of the ‘conditioned’ environments. The dependent measure of CPP is the difference in the amount of time spent in the environment associated with the presence of the putative reward vs. the environment paired with its absence. If the stimulus is rewarding, the mouse should spend more time in the environment that is paired with it. Among rodents, CPP tests have showed reward processes during different kinds of social encounters, including play (Calcagnetti & Schechter 1992, Douglas et al. 2004), sexual interactions (Camacho et al. 2004, Jenkins & Becker 2003), mother–infant bonding (Mattson et al. 2001, Weller et al. 2001) and even aggression (Martinez et al. 1995).

CPP experiments can also be helpful in determining whether mouse strains that express different USV repertoires experience different levels of social motivation. For instance, B6 mice show a strong place preference for bedding environments that have been associated with access to age-matched juveniles vs. beddings that are paired with social isolation (Panksepp & Lahvis 2007). These results suggest the possibility that the vocal repertoire of B6 mice may be associated with the degree of social reward; that there is a relationship between call pattern and various levels of social motivation. We can also gain insight to useful approaches for studying mouse USVs from insightful experiments of rat vocalizations. Rats emit high-frequency (50 kHz) calls when they anticipate a future reward or when they are experiencing the reward, and emit lower frequency (22 kHz) USVs under aversive conditions (Burgdorf et al. 2008, Knutson et al. 1999, Schwarting et al. 2007, Wöhr & Schwarting 2009). In addition to conditioning experiments, studies with operant tasks indicate that rats will self-administer playbacks of recordings of 50 kHZ ‘trill’ calls but not 22 kHz calls (Burgdorf et al. 2008). Similar kinds of evaluations might be useful for interpreting particular mouse USVs. It would be very useful to know whether specific call types associate with different CPP responses to specific drugs. Measures of mouse USVs have the potential to provide insight to real-time fluctuations in affect that we cannot obtain through CPP or fear conditioning testing. Experiments that take advantage of the differences in social motivation that accompany variations in the duration of social isolation or the time of day when social encounter occurs (Panksepp et al. 2008) could be used to discern how call patterns and responses to them change with social conditions.

Behavioral and systemic responses to vocalizations

Studies of animal acoustic communication utilize behavioral responses to recordings or ‘playbacks’ of particular kinds of animal calls to ascertain the meaning of various signals. In a series of experiments, we found that mice can become differentially fearful of a tone that is temporally associated with the vocalizations of a distressed mouse. This experiment involves a cue-conditioned fear learning paradigm (LeDoux et al. 1988). A standard fear conditioning procedure entails presenting an animal with a neutral stimulus, such as a tone (conditioned stimulus, CS), forward paired with an aversive stimulus, such as an electrical shock (unconditioned stimulus, UCS). Upon repeated administration of the paired CS–UCS, mice acquire a robust freezing behavior in response to presentation of the CS only. When B6 and BALB mice observe others undergo this fear conditioning procedure, only B6 mice are capable of learning from the distress of others. B6 also acquire the freezing response to the tone if they simply hear playbacks of the vocalizations of other distressed mice (played back 2-second recordings of the vocalizations of shocked mice) in association with the 30-second tone. Control experiments indicate that the BALB mice can hear the calls, because they respond to an environmental cue (the tone) of an overlapping frequency and energy level (approximately 85 dB). Importantly, heart rate deceleration is expressed in B6 when demonstrator mice undergo fear conditioning, but BALB mice do not show heart rate deceleration. Depression of heart rate can be associated with the empathic response to other's distress and this occurs among children more capable of detecting distress among others (Eisenberg et al. 2006). These studies indicate that fear can be perceived through vocal signals under the appropriate genetic background.

Brain activity in response to vocalizations

To assess the salience of different call types, researchers are beginning to examine freezing behavior, heart rate changes or other behavioral end-points in response to playbacks of auditory cues, such as distress vocalizations (Jeon et al. 2010, Knapska et al. 2006, Liu et al. 2006, Sadananda et al. 2008). For instance, variations in USV frequency modulation are sensitive to the age of the pups, with distinct influences on activity of the auditory cortex (Liu et al. 2003). In fact, experience appears to change neural processing of auditory stimuli; maternal auditory processing of pup USVs is distinct from that of naïve females (Liu et al. 2006, Liu & Schreiner 2007).

Future directions

  1. Top of page
  2. Abstract
  3. Prosody
  4. Genetics
  5. Future directions
  6. Summary
  7. References
  8. Acknowledgments

USV analysis

There are a variety of approaches to vocal analysis, ranging from classification schemes to principal components analysis to hidden-Markov approaches. Various software programs have been developed to serve different purposes and each has its own benefits. Software programs can be employed to directly measure USV s, such as Avisoft SASLabPro (Panksepp et al. 2007, Scattoni et al. 2008a, Wöhr et al. 2008), others are designed to dissociate calls based upon highly nuanced spectral features, such as continuity and entropy (Tchernichovski et al. 2001), or identify calls in low signal-to-noise conditions (Mellinger 2009). Investigators also utilize custom MATLAB software programs to analyze frequency modulation in mouse USVs (Liu et al. 2006). Pattern recognition in USV sequence is also aided by the application of advanced statistical approaches (Ren et al. 2009). In this regard, future studies will likely utilize a variety of software approaches to obtain different kinds of information from calls, ranging from simple classifications of relatively similar call types (Panksepp et al. 2007) to discriminations of quite subtle changes in frequency modulation within a single call type. Differences in affective content might then be discerned by use of playbacks in combination with measures of brain activity, heart rate changes, behavioral responses, or other parameters that allow us to make inferences to affective response.

Complex environments

The rodent has long been considered a mammalian substitute for the human patient (Beach 1950), but when concentrating on a single, increasingly exploited, lab species, it has to be taken in mind that such a species is the ‘ultimate’ product (a variable genetic pool fragmented in a variety of local populations) of species-specific phylogenetic and epigenetic processes. Laboratory mice have been strongly selected for husbandry in a small cage, where there are limited opportunities for temporal and geographic variations in food access, social encounter, temperature, diurnal rhythms, encounter with predators, or exploration. By contrast, feral mice, by definition strictly associated to human settlements (Bronson 1979), show highly variable adaptive strategies that are not promoted by standard cage housing. It is likely that laboratory mice are thus not raised to vocalize in response to the more complex variety of social-emotional states that accompany a multifaceted environment. In this regard, wild-derived mice provide potentially useful opportunities to dissociate USV patterns (Musolfa et al. 2010, Kalcounis-Rueppell et al. 2010). A truly, genuine, comparative approach should be based on a much broader range of mouse experiences and representative species: the search for functional and/or structural invariance could be a fruitful ‘zoo-semeiotic’ perspective (sensu Sebeok) (Sebeok 1972). Studies with feral mice, as well as other feral rodent species, might provide a high degree of fidelity between call type and affective state.

More substantive vocal behaviors might also be showed by naturalistic or semi-naturalistic settings (Alleva et al. 1994, Portfors 2007). While mouse vocalizations are sensitive to targeted loss-of-function mutations in genes involved in affect regulation, language development, social bonding and autism, we may gain greater insight to how these mutations modify call patterns by expanding mouse developmental histories. Genetic variants can be raised in complex settings to expand their emotional experience and vocal repertoire. Rearing could entail geographic and temporal variation, such as random temporal access to food rewards, such as peanut butter, and stressors, such as cat urine, within a complex three-dimensional environment. Mice raised in environments that incorporate rewarding and aversive stimuli, in the context of more complex social paradigms, might provide a useful model for delineating relationships between affective states and specific call types.

Summary

  1. Top of page
  2. Abstract
  3. Prosody
  4. Genetics
  5. Future directions
  6. Summary
  7. References
  8. Acknowledgments
  • Assessments of frequency modulation provide researchers with an ability to study USVs in a fashion that has translational relevance to prosodic deficits in mental illness. Measures of mouse USVs have the potential to provide a window to real-time changes in affective state that are not accessible by CPP or fear conditioning tests.
  • Inferences about call meaning should be considered within the context of ethologically relevant behaviors or traditional assessments of reward and aversion (e.g. conditioning experiments) and through the use of heart rate monitoring, assessments of brain activity, and other physiological indicators of affective response.
  • Mutant genetic models that are relevant to reward neurobiology, language, autism and social bonding can express abnormal USV patterns. Use of multiple statistical approaches for bioacoustic analysis, including clustering and hidden-Markov approaches, can help elucidate associations between USV structure and function.
  • By using wild-derived mice as rearing mice in more complex environments, researchers might also gain greater insight to the discrete functions, individual calls or call sequences within the diversity of the vocal repertoire.

References

  1. Top of page
  2. Abstract
  3. Prosody
  4. Genetics
  5. Future directions
  6. Summary
  7. References
  8. Acknowledgments

Acknowledgments

  1. Top of page
  2. Abstract
  3. Prosody
  4. Genetics
  5. Future directions
  6. Summary
  7. References
  8. Acknowledgments

The authors wish to thank Jan Van Santen and Lois Black of Oregon Health and Science University for communicating their insights to prosodic communication. We also want to acknowledge the Waisman Center at the University of Wisconsin for support (P30 HD03352). This work was supported by R01 funding from the National Institute of Drug Abuse (DA022543).