The Development of the Ability to Distinguish between
Genuine and Polite Laughter.
Is it Shaped by Social Learning?
UvA
Amber van der Wal
Masterthese Social Psychology (5665027) Supervisor: Disa Sauter PhD
Contents
Abstract 3
Introduction 5
Laughter 5
Different Types of Laughter 7
Genuine versus Simulated Laughter 9
Distinguishing Between Genuine and Simulated Laughter 11 Children’s Use and Understanding of Display Rules 12
Current study 14 Hypotheses 15 Method 18 Material 19 Participants 19 Procedure 20 Design 20 Dependent variables 21 Results 22
Development of the Ability to Distinguish Genuine from Polite Laughter 22
Recognition of Genuine versus Polite Laughter 26
Positivity Ratings 32
Discussion 34
Development of Discriminative Ability with Age 34
Role of Social Learning 35
Recognition of Genuine versus Polite Laughter 36
Perceived Positivity of Laughter 37
Limitations and Future Directions 38
Conclusions and Implications 42
Abstract
Many researchers have made a distinction between genuine, involuntary laughter and
simulated, voluntary laugher (e.g., Gervais & Wilson, 2005; Keltner & Bananno, 1997; Wild, Rodden, Groch, & Ruch, 2003). Whether and how capable people are in making this
distinction is poorly explored. We therefore examined adults’ and children’s ability to recognize and distinguish between polite (i.e., a subtype of simulated laughter) and genuine laughter.
The present study set out with three main aims. The first aim was to examine the development of the ability to recognize and distinguish between genuine and polite laughter in children and adults. The second aim was to determine whether frequent exposure to laughter improves these abilities (social learning). Since previous studies reported mixed results on the perceived positivity of different laughter types, the third aim was to examine differences in the positivity ratings of different types of laughter.
We conducted an experiment with 762 participants (497 children and 265 adults, age range 3-75 years), who were asked to listen to audio samples of genuine and polite laughter and judge these as either genuine or polite. In order to examine the possible differential development due to social learning, we included genuine and polite laughter samples from two different cultures (Dutch and Japanese).
Results showed that children from age six onwards were increasingly able to recognize and distinguish between genuine and polite laughter. In addition, from the time this ability emerged in children, they showed a better discriminative ability for Dutch laughter than for Japanese laughter and this ability kept improving until around the age of 40. However, discriminative ability for Japanese laughter did not improve in adulthood. These results provide support for the social learning hypothesis that holds that exposure to laughter
of laughter. Furthermore, examination of the positivity ratings of the different laughter types revealed that genuine and polite laughter were both rated as positive in valence, but genuine laughter was perceived by all participants as more positive than polite laugher. Together, these findings contribute to a better understanding of the (development of) perception, identification and recognition of different laughter types and provide suggestions for future research.
“Nothing is more serviceable than a laugh, in many of the exigencies of society. Many an argument finds its full answer in a guffaw, many a refusal is tempered by a kindly chuckle. Laughter of this sort is but a form of language. It is not easy to say where spontaneous laughter ends and controlled laughter begins.”
H.C. McComas, The origins of laughter, 1923
Introduction
Laughter is ubiquitous in human society. Nevertheless, the scientific study of laughter is still a relatively new area of research and many empirical questions remain unanswered. We aim to expand the small existing body of research by examining how the ability to recognize and distinguish between different types of laughter develops in children and adults and whether these abilities are shaped by social learning.
In this introduction will first be outlined (a) what laughter is, (b) the different types of laughter that exist, (c) the acoustic and perceptual differences between different types of laughter, and (d) perceivers’ ability to recognize and distinguish between these different types of laughter. Second, because to date no research has been conducted on this ability in
children, relevant findings from related areas of research on children’s social and cognitive development will be discussed.
Laughter
Laughter is a type of non-verbal vocalization (like crying and screaming) created by the vocal organs: the lungs, larynx, tongue, lips and occasionally the nose (Chafe, 2007). Air passes from the lungs through the larynx, out through the mouth (or sometimes the nose), creating a sound. This sound is typically modified in the larynx to make it more audible (giving it its volume).
Laughter is deeply grounded in human biology. Comparative and phylogenetic
analyses show that laughter has evolved from the relaxed, open-mouth play face exhibited by numerous primate species (Van Hooff, 1972) and associated pant-like vocalizations (Darwin, 1872; Gervais & Wilson, 2005; Provine, 2000). Laughter emerges early in development at around four months of age (Sroufe & Wunsch, 1972). However, reports of gelastic epilepsy (from the Greek gelos, meaning laughter) among neonates demonstrate that all necessary structures are already present and functional by birth (Ruch & Ekman, 2001). The
development of laughter does not require auditory experience to emerge in species-typical form, as has been shown in studies with congenitally deaf children and adults (Makagon, Funayama, & Owren, 2008; Scheiner, Hammerschmidt, Juergens, & Zwirner, 2006). The acoustic properties of laughter produced by deaf people are comparable to those of hearing people (Makagon et al., 2008). In addition, laughter is found in all cultures and virtually all individuals all over the world (Gervais & Wilson, 2005; Provine, 2000) and is cross-culturally recognized (Sauter, Eisner, Ekman, & Scott, 2010). Taken together, these findings suggest that human beings have a strong inherent predisposition for the development of laughter production and perception.
Although the basic act of laughter clearly has a biological foundation, this is not to say that laughter may not be shaped by experience as well (Gervais & Wilson, 2005). The things that elicit laughter, the contexts in which laughter emerges, and the expression, frequency, and intensity of laughter are all influenced by culture, that is, in accordance with display rules and varying norms and customs (Weisfeld, 1994). It has been found that although the sounds of basic emotions (including the sound of laughter, signaling amusement) are cross-culturally recognized, each group still recognizes sounds produced by members of their own culture better (Elfenbein & Ambady, 2002; Sauter et al., 2010), suggesting a role for culture-specific social learning.
Laughter has many functions in modern human society. These range from bonding and solidifying friendships, easing tensions and competition, and synchronizing the emotions and behaviors of a group, to undoing negative emotion and stress and promoting emotional stability and physiological and psychological health (Gervais & Wilson, 2005; Keltner & Bonanno, 1997; Mora-Ripoll, 2011; Provine, 2000).
Laughter is also inherently social. People laugh about 30 times more in the presence of others than when they are alone (Provine, 2000), and laughter is virtually non-existent in solitude (Chapman, 1983; Provine & Fischer, 1989). As Marijuan and Navarro (2010) state, “laughter is one of the most complex behaviors exhibited by humans. It integrates the innate and the cultural, the emotional and the cognitive, the individual and the social” (p. 10).
Different Types of Laughter
Although laughter is largely a stereotyped vocalization, explaining its universally
recognizable sound (Provine, 2000), many different types of laughter can be distinguished. An analysis of different dictionaries, historic literature and online corpora has shown that more than 200 different types of laughter can be differentiated on a linguistic basis (Drack, Huber, & Ruch, 2009). Some of those categories are purely evaluative or descriptive (e.g., roaring laughter, or wild laughter). Others contain affective information by referring to the emotion and motivation of the person laughing (Drack et al., 2009).
Different categorizations of laughter have also been described in the research literature (Mora-Ripoll, 2011). For example, a distinction has been made between humorous and non-humorous laughter (Chafe, 2007; Chapman, 1983). Laughter can occur in response to a humorous event. However, often things said just prior to laughter are not humorous
(Chapman, 1983; Provine, 2000). Non-humorous laughter is in many instances triggered by situations in which the laughter can mitigate unpleasantness (Chafe, 2007), for example in
case of embarrassment or criticism. Non-humorous laughter can also express negative feelings and attitudes such as contempt, sadness and anger (Devillers & Vidrascu, 2007).
Laughter has also been categorized according to valence: positive (amused, joy, sympathetic, polite, relief laugh), negative (disappointment, embarrassed, stressed laugh) or ambiguous (comment laugh, ironical laugh) (Devillers & Vidrascu, 2007). An issue with this categorization is that laughter can also differ in degree of volitional control (Ruch & Ekman, 2001). For instance, amused laughter and polite laughter both belong to Devillers’ and Vidrascu’s category of positive laughter, but these differ greatly in the sense that the first is genuine and the latter is non-genuine.
Many researchers therefore differentiate different kinds of laughter by how they are elicited (Scott, Lavan, Chen, & McGettigan, 2014) and hence have made a distinction between laughter that is voluntary, that is, controlled, as compared to laughter that is involuntary, that is, uncontrolled (e.g., Gervais & Wilson, 2005; Keltner & Bananno, 1997; Wild, Rodden, Groch, & Ruch, 2003). In the next section, these two types of laughter will be discussed in further detail, but first a word on terminology is in order.
Spontaneous, involuntary laughter has also been called “genuine” (Lavan, Lima, Harvey, Scott, & McGettigan, 2014; Reysen, 2006), “mirthful” (Miller & Fry, 2009; Tanaka & Campbell, 2011), “felt” (LaFrance, 1983), “social” (Scott et al., 2014; Vettin & Todt, 2004), “Duchenne” (Gervais & Wilson, 2005; Keltner & Bonanno, 1997), “ humorous” (Chafe, 2007), “voiced” (Hudenko, Stone, & Bachorowski, 2009) or “authentic” (McGettigan, Walsh, Jessop, Agnew, Sauter, & Scott, 2013) laughter. As their antonyms, terms such as “simulated” (Mora-Ripoll, 2011), “posed” (Galvan, Manangan, Sanchez, Wong, & Cu, 2011; Lavan et al., 2014), “polite” (Tanaka & Campbell, 2011), “feigned” (LaFrance, 1983), “fake” (Reysen, 2006), “forced” (Foley, Matheis, & Schaefer, 2002), “contrived” (Ruch & Ekman, 2001), “non-Duchenne” (Gervais & Wilson; Keltner & Bonanno, 1997), “unvoiced”
(Hudenko et al., 2009), “volitional” (Bryant & Aktipis, 2014) or “non-humorous” (Chafe, 2007) have been used.
Many of the terms used for involuntary laughter are closely semantically related and it is more a matter of taste which term one prefers. From here on, the term “genuine” will be used to refer to involuntary laughter, because this term reflects the emotional experience that one has when laughing, rather than a more behavioral description like spontaneous or
uncontrolled.
Many of the terms used for voluntary laughter are not so closely semantically related. Some are broad umbrella terms (e.g., simulated and posed) and some are specific forms of voluntary laughter (e.g., fake laughter, polite laughter). Others, such as (un)voiced, (non)-Duchenne, and (non-)humorous, refer to characteristics of laughter that are often co-occurring with, but not the defining feature of this type of laughter. From here on, as the antonym of genuine, the term “simulated” will be used. This term encompasses spontaneous as well as enacted simulated laughter and different subtypes such as polite and fake laughter. We have decided to use this term because research on this topic is scarce, and using this broad umbrella term enables us to incorporate results from all studies that have been conducted on this topic.
Genuine versus Simulated Laughter
Genuine laughter is the behavioral component of a spontaneous, positive emotional reaction (often called “mirth” or “amusement”) to a specific stimulus (Gervais & Wilson, 2005). In contrast, simulated laughter is thought to be triggered by oneself at will (self-induced), not elicited by humor, and not necessarily accompanied by an emotional experience (Gervais & Wilson, 2005). Genuine and simulated laughter also appear to be acoustically different (see Devillers & Vidrascu, 2007; Kipper & Todt, 2001; 2003; Ruch & Ekman, 2001; Tanaka &
Campbell, 2011; Vettin & Todt, 2004), although it is not yet established which specific acoustic features constitute genuine and simulated laughter.
Gervais and Wilson (2005) have theorized that it is only the involuntary, genuine type of laughter that has the before mentioned ancient phylogenetic roots. They posit that “genuine laughter originally evolved in the context of proto humor and functioned as a medium of joyful emotional contagion” (p. 418), whereas simulated laughter evolved much later, allowing people to strategically influence others.
This idea is supported by neuroscientific findings. Wild et al. (2003) have elucidated two partially dissociable neural pathways underlying the two different types of laughter. Additional support comes from a study using PET (Iwase, Ouchi, Okada, Yokoyama,
Nobezawa, Yoshikawa, & Watanabe, 2002), which demonstrated distinct neural substrates of emotional and volitional smiling. Recently, McGettigan et al. (2013), using fMRI, have shown that simulated and genuine laughter engage different neurocognitive systems.
So, genuine and simulated laughter appear to be acoustically different and produced by different underlying neural control systems. Whether they are also perceived differently in terms of positivity remains unclear. Several theorists have suggested that positive affect is only elicited by genuine laughter (Bachorowski & Owren, 2001; Gervais & Wilson, 2005; Ruch & Ekman, 2001). However, other researchers have claimed that simulated laughter can lead to an increase in positive affect (Foley, Matheis & Schaefer, 2002; Neuhoff & Schaefer, 2002). In addition, McGettigan et al. (2013) have found that both types of laughter are judged as positive in valence, but genuine laughter is perceived as more positive than simulated laughter.
Distinguishing Between Genuine and Simulated Laughter
Simulated laughter seems to constitute the majority of laughter in conversations (Chafe, 2007; Gervais & Wilson, 2005; Provine, 2000; Tanaka & Campbell, 2011). Such laughter often functions to smooth interactions and promote positive feelings between people (Vettin & Todt, 2004), but people can also learn to use simulated laughter in a strategic and/or deceptive manner.
Simulated laughter has even been observed in infants, as reported by many parents, who describe this type of laughter as very distinctive from genuine laughter (Reddy, 2007; 2008). However, it should be noted that this was reported by the infants’ parents, not by an objective observer, and that Reddy is the only author who has described the distinctiveness of genuine and simulated laughter in children. Nevertheless, although little is known about children’s laughter in general, the findings reported by Reddy support results from other studies that have focused on psychological functions of children’s laughter (Chapman, 1973; 1974; 1975a; 1975b; 1983; Chapman & Chapman, 1974; Chapman & Wright, 1976; Sarra & Otta, 2001), that already in childhood laughter is more than just a pure expression of mirth.
A few studies have examined adults’ ability to distinguish genuine from simulated laughter. McGettigan et al. (2013) showed that in general, participants are able to distinguish between genuine and simulated laughter with a high degree of accuracy, on average 82.5% correct. There were however, considerable individual differences between participants (ranging from 69% to 93% accuracy). Similar results were reported by Bryant and Aktipis (2014), who found a lower, but still well above chance, degree of accuracy, with participants being on average 67% correct. Even lower accuracy levels were found by Reysen (2006) (mean correct response was 56.5%), where participants could detect the difference between fake and genuine laughter at a rate very slightly (but statistically significantly) better than chance.
In sum, genuine and simulated laughter seem to be perceptibly and acoustically different and are produced by different underlying neural control systems. Nevertheless, adults (at least in experimental studies) still get fooled rather frequently, although the level of discriminative ability differs greatly between studies (Bryant & Aktipis, 2014; McGettigan et al., 2013; Reysen, 2006). Taken together, these findings raise the question whether people perform with the same level of accuracy across the lifespan or whether the ability to
distinguish between different types of laughter may be age-dependent. Therefore, the aim of the current study is to compare adults’ and children’s ability to distinguish between different types of laughter, in particular between genuine and polite laughter (a subtype of simulated laughter).
To date, no research has been conducted on children’s ability to distinguish between genuine and simulated laughter. Therefore, findings from related areas of research, that is, children’s use and understanding of (prosocial) display rules, will be discussed in the next section, on the basis of which hypotheses will be derived.
Children’s Use and Understanding of Display Rules
Display rules are a set of culture-specific norms for emotional expression that dictate when to express, manage or modify one’s emotions (Ekman & Friesen, 1975). People may control the display of their emotions for various reasons (Gnepp & Hess, 1986; Gosselin, Warren, & Diotte, 2002). Two broad categories of motivation have been delineated: self-protective and prosocial (Gnepp & Hess, 1986; Josephs, 1994). Self-protective display rules are primarily employed in order to avoid negative consequences for oneself and/or preserve one’s self-esteem (Gosselin et al., 2002; Josephs, 1994). Prosocial display rules serve to protect the feelings of others and maintain harmonious relationships (Gosselin et al., 2002; Josephs, 1994).
Prosocial display rules are often employed in politeness situations, for example
through speech (e.g., prosocial “white” lies) or facial expressions (e.g., non-enjoyment smiles, Del Giudice & Colle, 2007; Gosselin, Perron, & Maassarani, 2010). We would argue that polite laughter is just like a white lie or a non-enjoyment smile a form of prosocial behavior, only expressed through a different channel (non-verbal vocalization instead of a verbal vocalization or facial expression). Because of the lack of research on polite laughter, we will discuss findings from studies on these other forms of prosocial behavior in children, in order to formulate hypotheses for the current study.
Prosocial lies (such as telling someone that the meal he cooked tastes good, when in fact it does not) are, just like polite laughter, common in everyday adult interactions (DePaulo & Kashy, 1998). Studies that have examined children’s prosocial lie-telling behaviors have shown that on average, the majority of child participants, ranging from preschoolers to late elementary school children, tell white lies (Popliger, Talwar, & Crossman, 2011; Talwar, Murphy, & Lee, 2007; Xu, Boa, Fu, Talwar, & Lee, 2010), but children are increasingly more likely to lie during the elementary school years (Broomfield, Robinson, & Robinson, 2002; Popliger et al., 2011; Xu et al., 2010).
Prosocial lies are seen by most adults as socially acceptable (DePaulo & Jordan, 1982). However, preschoolers evaluate lying in politeness situations negatively, although they do perceive white lies as less negative than antisocial (self-protective) lies (Bussey, 1999). Between the preschool and elementary school years prosocial lies become perceived as less negative (Popliger et al., 2011; Xu et al., 2010). In the elementary school years prosocial lies are perceived as more neutral, whereas antisocial lies are always rated as negative (Heyman & Sweet, 2009). In adolescence prosocial lies start to be perceived as positive (Lee, 2013).
Studies on children’s understanding and use of prosocial display rules in facial
of six children show an explicit, articulated understanding of display rule use. Children are able to describe situations in which they smiled because they were happy as well as situations in which they were not happy, but smiled in order to look happy (Gosselin et al., 2010). They can also explain why it is appropriate to smile in certain social situations even if no happiness, or a related positive emotion (e.g., amusement), is actually experienced (Gosselin et al., 2010).
Children's understanding of how and when to regulate the display of emotion is developed through cognitive maturity as well as social experience (Talwar & Crossman, 2011). Children gradually develop a Theory of Mind (ToM), which refers to the cognitive ability to attribute mental states (thoughts, knowledge, desires, emotions, beliefs, intentions) to oneself and others and to understand that these may differ from each other (Sodian & Kristen, 2009). In addition, children become increasingly socialized to the norms of
politeness, a process that is facilitated by children’s interactions with parents and peers who either implicitly or explicitly teach them the rules of communication and in what context which rule of communication applies (Talwar & Crossman, 2011). As children get older they engage in situations in which they can practice their own abilities and observe how others use display rules (Talwar & Crossman, 2011).
Current study
The first aim of this study is to examine (the development of) children and adults’ ability to recognize and distinguish between genuine and polite laughter. The ability to recognize and distinguish between these two types of laughter will be measured by asking children and adults of different ages to classify audio samples of genuine and polite laughter in a two-way forced choice (genuine or polite). A second aim is to study the influence of social experience (exposure) on the ability to distinguish between genuine and polite laughter. Therefore,
laughter samples from two different cultures (one’s own culture and a foreign culture) will be included. A third aim is to investigate whether laugher type influences the perceived positivity of the laughter. For this reason, the participants will also be asked to rate the perceived
positivity of the genuine and polite laughter samples from their own and a foreign culture.
Hypotheses
Hypothesis 1). Children younger than five years of age do not perform above chance level in distinguishing between genuine and polite laughter because they lack the necessary cognitive skills.
We assume that the ability to distinguish between genuine (real) and polite (apparent) laughter develops in a way that is comparable to the development of the ability to distinguish between real and apparent emotions. Most studies have corroborated that this ability emerges
somewhere between four and six years of age (e.g., Josephs, 1994; Misailidi, 2006; Sidera, Amadó, & Serrat, 2013; Sidera, Serrat, Rostan, & Sanz-Torrent, 2011), together with the development of first-order false belief understanding (a specific subskill of ToM). The general consensus seems to be that four-year-olds begin to have a rudimentary, implicit understanding that pretend emotions are not real, and that around the age of six children’s ability to
distinguish real from apparent emotions becomes systematic and their knowledge of it explicit (Sidera et al., 2013). Since four-year-olds only seem to begin understand this distinction, we expect that children younger than five years of age are not able to distinguish between
genuine and polite laughter above chance level. Second, based on the studies discussed above, we expect that perhaps as early as five years of age and certainly from six years of age,
Hypothesis 2). The ability to distinguish between genuine and polite laughter improves with age.
We expect that, from the time children perform above chance level in distinguishing genuine from polite laughter, they will show an increase in discriminative ability with age, due to cognitive development and social learning. Previous research has shown that the ability to distinguish real from apparent emotions improves at least until the age of eleven years (Pons, Harris, & De Rosnay, 2004). This improvement may even continue well into adulthood. Scott et al. (in prep) found that the ability to discriminate between genuine and polite laughter keeps improving until people are in their thirties. So, we expect that overall, adults’ discriminative ability is better than children’s and that older children perform better than younger children.
Hypothesis 2a). There is an accelerated increase in the ability to distinguish between genuine and polite laughter between the ages of four and six.
Apart from the expected general improvement (a linear trend) with age, we expect to see an accelerated increase (a “bump”) in the ability to distinguish genuine from polite laughter between the ages of four and six, due to emerging cognitive abilities. This is based on the fact that this accelerated increase between the ages of four and six has also been found for the ability to distinguish real from apparent emotions (Josephs, 1994; Joshi & MacLean, 1994; Misailidi, 2006; Sidera et al., 2013).
Hypothesis 2b). The ability to distinguish between genuine and polite laughter improves with age due to exposure.
We also expect an increase in the ability to discriminate between genuine and polite laughter due to exposure. People will be more frequently exposed to laughter (genuine or polite) expressed by members from their own culture than by people from a different culture. This
means that a differential effect is expected in the ability to distinguish between genuine and polite laughter from one’s own culture (in this case Dutch) and from a foreign culture (in this study Japanese). This is supported by related studies (Elfenbein & Ambady, 2002; Sauter et al., 2010), who found that there is an in-group advantage for the recognition of emotional vocalizations in general, meaning that they are better recognized when produced by members of their own culture than when produced by another culture. Sauter et al. (2010) conclude that these cultural variations are acquired through social learning. The expectation is therefore that children as well as adults will show a better discriminative ability for laughter from their own culture, because they have been exposed to it more often.
Hypothesis 3a). Genuine laughter is better recognized than polite laughter.
Bryant and Aktipis (2014) found a significant difference between adult participants’ ability to correctly judge genuine laughter (71%) and simulated laughter (63%). In addition, Scott et al. (in prep) found that children as well as adults reached higher levels of accuracy in recognizing genuine laughter compared to polite laughter. Due to the ancient and deep biological
foundations of genuine laughter, it is predicted that children, from the moment they start distinguishing between genuine and polite laughter, are better at recognizing genuine laughter than polite laughter, and that this higher recognition level of genuine laughter stays present in adulthood.
Hypothesis 3b). Genuine laughter from both one’s own and a foreign culture is better recognized than polite laughter.
Sauter et al. (2010) found that the sound of laughter signaling genuine amusement is cross-culturally recognized. It is expected that due to the deep biological roots of genuine laughter
and the presumably greater influence of culture on polite laughter, genuine laughter from both cultures will be recognized better than polite laughter by children as well as adults.
Hypothesis 4a). Genuine laughter is perceived as more positive than polite laughter.
It is expected that, in accordance with the findings of McGettigan et al. (2013), both genuine and polite laughter will be judged as positive in valence. In addition, a difference in positivity-scores between genuine and polite laughter is expected (Bachorowski & Owren, 2001;
Gervais & Wilson, 2005; McGettigan et al., 2013; Ruch & Ekman, 2001), in that genuine laughter will be rated as more positive than polite laughter. We expect that younger children will rate polite laughter as less positive than older children, just as white lies become less negatively perceived as children grow older.
Hypothesis 4b). Laughter from one’s own culture is perceived as more positive than laughter from a foreign culture.
According to Zajonc’s (1968) mere exposure theory, people tend to develop a preference for things merely because they are familiar with them (often called the familiarity principle). This effect has been demonstrated across many categories, including words, Chinese characters, paintings, pictures of faces, geometric figures, and sounds (Zajonc, 2001). It is therefore expected, since people are more frequently exposed to laughter from their own culture than from a foreign culture, that laughter from their own culture will be rated as more positive.
Method
This study consists of secondary data analysis. The data used in this study were obtained as part of the research project “Is Dat Echt Lachen?”, conducted in April and May 2014 at Science Center Nemo, a museum in Amsterdam. Children as well as adults participated.
Ethical approval was obtained from the Departmental Ethics Committee at the Department of Psychology, University of Amsterdam. Informed consent was obtained from all adult
participants and from parents or guardians for all participants younger than 18 years.
Material
The stimulus material consisted of 16 laughter clips that were balanced in terms of culture (Dutch versus Japanese laughs), sex (female versus male laughs), and genuineness (genuine versus polite laughs). All factors were crossed. Four different Dutch speakers and five different Japanese speakers were included in the sample. All sounds were downsampled to 44.1 kHz, converted to mono and scaled to a peak amplitude of 0.95 Pa.
The laughter exemplars were selected from a sample of 795 laughs tested previously in a pilot study. In this pilot study, recordings of genuine laughter were gathered through letting Dutch and Japanese participants watch laughter-inducing videos (that they selected
themselves). Recordings of polite laughter were collected through telling the participants bad jokes at which they had been instructed to laugh politely. The 16 laughs that made the final selection were the ones that were correctly judged as being most genuine or polite, and as sounding most authentic (i.e., not clearly exaggerated).
Participants
A total of 976 participants were tested. From these, 762 were included in the final sample (214 participants failed either to complete all trials, to fill in their age or sex on the consent form, or were excluded because they were not Dutch). Of these 762 participants, 497 were children, that is, 18 years or younger (249 girls, age range 3-18, with a mean age of 9.63 years, SD = 2.65) and 265 were adults (161 females, age range 19-75, with a mean age of 42.98 years, SD = 9.62). See Table 1 for a distribution across different age levels.
Procedure
All visitors of Nemo could participate in the study, including people from countries other than the Netherlands. They were however excluded from the final sample in the current analysis, because of possible cultural influences. There were six experimenters in total, with a
minimum of two present at any one time. There were four testing stations, so up to four participants could be tested at the same time.
Participants received written instructions that explained that they would hear genuine (“occurs when someone finds something really funny”) and polite (“occurs when someone laughs even though they do not consider something to be funny”) laughter from either Dutch or foreign people. In addition, they were told that laughs can differ in positivity: some laughs may sound very positive, while others may sound just a little positive. Then, all 16 laughter trials were played through headphones in a random order. Participants were sitting in front of a computer screen on which the response alternatives were displayed. For each trial,
participants were asked to judge (a) whether the laugh was genuine or polite, (b) whether it was from a Dutch person or a foreigner, and (c) how positive the laugh sounded. Judgment order was constant for each participant, but counter-balanced across participants. For questions a and b, order of response alternatives was counter-balanced across participants.
Design
This study used a 2x2 experimental design with the factors “Genuineness” (genuine or polite laughter) and “Culture” (Dutch or Japanese laughter) as the independent variables. For each of these conditions (“Genuine Dutch”; “Genuine Japanese”; “Polite Dutch”; “Polite
Accuracy”, and “Perceived Positivity”. For each of these three measures separate scores were calculated.
Dependent variables.
The positivity of the laughter was rated on a seven-point Likert scale (lowest score was 1 and the highest score was 7), ranging from “A little bit positive”, accompanied by a smiley face of low intensity, to “Very positive”, accompanied by a high intensity smiley face. For the
measure “Perceived Positivity” average scores were calculated for the four trials for each of the four conditions.
The measures of Genuineness and Culture were dichotomous, because in each trial the participant made a judgment of whether the laughter was genuine or polite (Genuineness) and Dutch or foreign (Culture). Using these outcomes, two different types of scores were
calculated: a d’ score and a recognition score.
The first score, d’, is a discriminability index, that reflects a person’s ability to discriminate between signal and noise, or in this case between genuine and polite or Dutch and foreign. The formula used to calculate d’ is Z(hit rate) – Z(false alarm rate) (e.g., Stanislaw & Todorov, 1999). For each participant, the proportional hit rate and proportional false alarm rate on the four trials was calculated (so, the lowest score is 0 meaning only incorrect responses and the highest score is 1 meaning perfect performance). These were then transformed into z-scores (in order to do so, all scores of 0 and 1 were replaced by 0.125 and 0.875 respectively, following the guidelines provided by e.g., Stanislaw & Todorov, 1999). This resulted in a possible score range of -2.30 through 2.30, in which the lowest score
indicates no hits and four false alarms and the highest score four hits and no false alarms. This score was used to examine the (differential) development of the ability to distinguish between (Dutch and Japanese) genuine and polite laughter (hypotheses 1 through 2b).
However, there were also specific expectations formulated about differences between the recognition of genuine and polite laughter (hypotheses 3a and 3b). This could not be examined with the discriminability index, since it only reflects participants' ability to distinguish between – and not the (possibly) differential recognition of - genuine and polite laughter. Therefore, a second type of index was computed using the following formula: hits/ (hits + misses + false alarms). This approach was chosen over using simply the raw
proportion correct scores, because the current approach provides a more precise measure of level of recognition, since the number of false alarms is also taken into account. This score has a range of 0 to 1, in which a score of zero means no hits, and a score of 1 means four hits, no misses and no false alarms. This index was calculated for all four conditions (“Genuine Dutch”; “Genuine Japanese”; “Polite Dutch”; “Polite Japanese”).
Unfortunately, participants were not equally distributed across ages. Some age levels were not represented at all, and others consisted of just a few participants. Therefore, for some analyses and the graphs and tables, age categories were formed that contained a substantial number (range of 36 to 88 participants per age category) of participants, in order to provide more reliable results.
Results
Development of the Ability to Distinguish between Genuine and Polite Laughter For hypotheses 1 through 2a, only the scores for Dutch laughter were needed to test the original hypotheses. We have however, in an exploratory fashion, run the same analyses for Japanese laughter. The results of these analyses are reported as well.
1) Children younger than five years of age do not perform above chance level in distinguishing between genuine and polite laughter.
Since there were only four participants under the age of five in the sample, the original hypothesis could not be tested. Therefore, five-year-olds were included in this analysis as well. Since a Shapiro-Wilk’s test (p < .05) showed violation of the normality assumptions of the data for children in this age range, we conducted a nonparametric Wilcoxon signed rank test that compared the median of the d’-scores on the ability to distinguish laughter types (hereafter: d'-scores) of three- through five-year-olds to zero (chance performance). The median of the d’-scores for Dutch laughter was not significantly different from zero (z = .68, p = .50). Exploratory analyses revealed that the same result was found for Japanese laughter (z = .45, p = .65).
A follow-up nonparametric Wilcoxon signed rank test revealed that six-year-old children (whose scores also showed violation of the normality assumptions) were in fact able to distinguish between genuine and polite laughter above chance level, (z = 3.45, p = .001). Again, the same result was found for Japanese laughter (z = 3.13, p = .002).
This offers support for the hypothesis that children younger than five years of age cannot distinguish between genuine and polite laughter. This seems to be the case for five-year-olds as well. From six years of age, children are able to distinguish between genuine and polite laughter above chance level.
2) The ability to distinguish between genuine and polite laughter improves with age.
A linear regression with the d’-scores for Dutch laughter as the dependent variable and age as the independent variable established that age positively predicts how well participants could distinguish between genuine and polite laughter, F(1,760) = 62.09, b* = .28, p < .001, see Figure 1. Age accounted for 7.6% of the variability in participants’ discriminative ability.
To see whether the improvement took place in childhood and/or in adulthood, we ran the same analysis separately in the sample of children (age range 3-18) and in the sample of adults (age range 19-75). A linear regression indicated that for children, age was a positive predictor, with older children being better able to distinguish these forms of laughter than younger children, F(1,496) = 56.89, b* =.32, p < .001 (age accounted for 10.3% of the
variability in participants’ level of discriminative ability). However, age was not a significant predictor for adults, F(1,263) = .69, p = .41. This indicates that the increase in discriminative ability with age happens primarily during childhood.
Figure 1. Discriminative ability for Dutch and Japanese laughter per age category.
2a). There is an accelerated increase in the ability to distinguish between genuine and polite laughter between the ages of four and six.
The six-year-olds’ level of discriminability could not be compared directly with that of the four-year-olds (since there were only four four-year-olds in the sample). However, a
comparison between the six-year-olds and the five-year-olds did show an accelerated increase in discriminative ability: the difference in level of discriminability between these two age groups was much bigger (mean difference score was .46) than for example between six- and seven-year-olds (mean difference score was .05). A similar increase in discriminative ability
0 0,2 0,4 0,6 0,8 1 1,2 1,4 1,6 3-6 7 8 9 10 11-12 13-18 19-35 36-40 41-45 46-50 51-75 d ´-sco re Age in years Dutch Japanese
was found for Japanese laughter (mean difference score between five- and six-year-olds was .40 and between six- and seven-year-olds .10). Since there were only 13 five-year-olds in the sample these results should be treated with some caution, but they offer initial support for this hypothesis.
2b). The ability to distinguish between genuine and polite laughter improves with age due to exposure.
In order to test this hypothesis a one-way repeated measures ANOVA was conducted with Stimulus Culture (Dutch versus Japanese) as a within-subject variable and participant’s age and sex as between-subject variables. Previous studies on related areas of research have reported mixed results on sex differences, some found none (e.g., Jenkins & Astington, 1996), some found an advantage for males (e.g., Gosselin et al., 2002) and others an advantage for females (e.g., Banerjee, 1997). Therefore, participant’s sex was added to explore whether there were any sex differences. The ANOVA was done for adults and children separately. First, the results for children will be discussed, followed by the results for adults.
Children. A main effect of Culture was found, F(1,496) = 24.40, p < .001, 95% CI [0.15, 0.34], partial η² = .05. Children’s discriminative ability was better when the laughter was Dutch (M = .97, SD = .84) as compared to Japanese (M = .73, SD = .89). There was also a main effect of age, F(1,495) = 77.49, p < .001, partial η² = .14. As can be seen in Figure 1, older children performed better than younger children. There was no interaction effect with age, which means that older children were better than younger children in distinguishing between genuine and polite laughter both for Dutch and Japanese laughter. There was also a main effect of sex, F(1,495) = 7.34, p < .01, 95% CI [0.04, 0.28], partial η² = .02. Girls displayed a higher level of discriminative ability (M = .93, SD = .85) than boys (M = .77, SD = .87).
Adults. A main effect of Culture was found, F(1,264) = 17.66, p < .001, 95% CI [0.13,0.35], partial η² = .06. Participants’ discriminative ability was significantly better when the laughter was Dutch (M = 1.39, SD = .65) than when it was Japanese (M = 1.16, SD = .72). There was no main effect of age, which is in line with the previous findings from the
regression analysis. The analysis did reveal a significant but weak interaction between Culture and age, F(1,263) = 4.41, p < .05, partial η² = .02. As can be seen in Figure 1 this interaction was caused by a continued improvement in discriminative ability with age for Dutch laughter and a stagnation in discriminative ability with age for Japanese laughter. Usually, over the years, adults will be frequently exposed to Dutch laughter and not to Japanese laughter. This corresponds with the finding that adults improved in their ability to distinguish between Dutch laughter types, but did not in their ability to distinguish between Japanese laughter types. There was no main effect or interaction effect involving participant sex, meaning that the sex differences that were present in childhood (i.e., a female advantage) had disappeared.
Recognition of Genuine versus Polite Laughter
Before hypotheses regarding the (differential) recognition of genuine and polite laughter could be tested, the existence of a possible response bias had to be taken into account. A response bias is the tendency to choose one alternative over the other, regardless of
recognition accuracy, which could obscure or exaggerate any true differences in recognition. The response bias was calculated for Dutch and Japanese laughter separately using the following formula from Stanislaw and Todorov (1999): -0.5*((Z(hit rate) + Z(false alarm rate)). The response bias had a possible range of -1.15 to 1.15, in which the lowest score would indicate that genuine was always chosen as the correct answer and the highest score would mean that polite was always chosen as the correct answer. A score of zero means that
there is no response bias present. Mean response bias scores for each age category can be found in Table 1.
Although close to zero, a one-sample t-test revealed that both the response bias for Dutch laughter (M = -.14, SD = .38, t(761) = -9.98, p < .001) and for Japanese laughter (M = .09, SD = .46, t(761) = 5.19, p < .001) differed significantly from zero. One-sample t-tests were also conducted for children and adults separately (with a Bonferroni correction for multiple comparisons). Children’s response bias was significant for Dutch laughter, (M = -.15, SD = .39, t(496) = -8.59, p < .001) and for Japanese laughter (M = .15, SD = .46, t(496) = 7.19, p < .001). The mean response bias scores were in opposite direction, which means that children were slightly more inclined to choose genuine when the laughter was Dutch (mean response bias had a negative value) and slightly more inclined to choose polite when the laughter was Japanese (mean response bias had a positive value). Adults’ response bias was only significant for Dutch laughter (M = -.10, SD = .33, t(264) = -5.11, p < .001).
Adults’ mean response bias score was in the same direction as children’s response bias score for Dutch laughter, meaning that both were slightly more inclined to choose genuine than polite. In conclusion, some response biases were found that should be kept in mind during the interpretation of the results of the following tests. These were however quite small with maximum scores of -.24 to .21, set against the possible maximum scores of -1.15 to 1.15.
Table 1. Mean Response Bias and Standard Deviation for Dutch and Japanese laughter per Age Category.
Age N Response bias
Dutch SD Response bias Japanese SD 3-6 48 -.10 .55 .19 .52 7 62 -.14 .43 .21 .50 8 63 -.14 .38 .21 .47 9 88 -.17 .39 .16 .46 10 76 -.24 .37 .12 .47 11-12 87 -.14 .37 .08 .42 13-18 73 -.10 .30 .10 .38 19-35 36 -.03 .44 .02 .40 36-40 66 -.14 .28 -.06 .45 41-45 83 -.11 .28 -.12 .37 46-50 40 -.11 .43 .11 .49 51-75 40 -.11 .32 .02 .49
3a). Genuine laughter is better recognized than polite laughter.
With this in mind, the hypothesis that genuine Dutch laughter is better recognized than polite Dutch laughter was tested (for which the second index, i.e., recognition performance, was used). A one-way repeated measures ANOVA was conducted with Genuineness (Genuine versus Polite) as the within-subject variable and participant’s sex as the between-subject variables. Participant’s sex was added to explore whether there were any sex differences. The ANOVA was done for adults and children separately.
Children A main effect of Genuineness was found, F(1,495) = 69.56, p < .001, 95% CI [0.04, 0.06], partial η² = .12. Genuine laughter (M = .58, SD = .21) was better recognized than polite laughter (M = .53, SD = .24). There was also a main effect of participant’s sex, F(1,495) = 8.78, p < .005, 95% CI [0.02, 0.09], partial η² = .02. Girls (M = .58, SD = .23) were better than boys (M = .53, SD = .22) at recognizing Dutch laughter.
Adults. A main effect of Genuineness was found, F(1,263) = 15.48, p < .001, 95% CI [0.01, 0.04], partial η² = .06. Genuine laughter (M = .68, SD = .20) was better recognized than polite laughter (M = .65, SD = .20). No sex differences were found.
In addition, the prediction that children are, from the moment they start discriminating between genuine and polite laughter, better at recognizing genuine laughter than polite
laughter, was also tested. As discussed earlier, from six years of age, children were able to discriminate between genuine and polite laughter above chance level. Six-year-old children were better at recognizing genuine laughter (M = .50, SD = .21) than polite laughter (M = .46, SD = .26), but this difference was not significant, t(29) = 1.06, p = .30. However, the
difference in level of recognition between genuine laughter (M = .46, SD = .25) and polite laughter (M = .50, SD = .23) of seven-year-olds was significant, t(61) = 2.37, p < .05. This difference may however be due to sample size, as the mean difference score for six- and seven-year-olds was the same, but there were double the amount of seven-year-olds (30 versus 62) in the sample.
Taken together, these findings offer support for the hypothesis that children, from the moment they start distinguishing between genuine and polite laughter, are better at
recognizing genuine laughter than polite laughter, and that this higher recognition level of genuine laughter stays present in adulthood.
3b). Genuine laughter from both one’s own and a foreign culture is better recognized than polite laughter.
To test this hypothesis a two-way repeated measures ANOVA was conducted with Culture (Dutch versus Foreign) and Genuineness (Genuine versus Polite) as the within-subject variables and participant’s sex as a between-subjects factor. This was done for children and adults separately. In Figures 2 and 3, children’s and adults’ mean recognition scores for the four types of laughter at different age levels are displayed.
Children. There was a main effect of Culture, F(1,495) = 21.15, p < .001, 95% CI [0.13, 0.17], partial η² = .04. Post hoc analysis with a Bonferroni adjustment showed that Dutch laughter was better recognized (M = .56, SD = .22) than Japanese laughter (M = .50, SD = .23). There was no main effect of Genuineness, F(1,495) = .16, p = .69, but there was an interaction effect between Culture and Genuineness, F(1,496) = 139.83, p < .001, partial η² = .22 (see Figure 2). Post hoc analysis with a Bonferroni adjustment showed that (as we already found in 3a) genuine Dutch laughter was better recognized than polite Dutch laughter. It was the other way around for Japanese laughter: polite Japanese laughter (M = .52, SD = .22) was better recognized than genuine Japanese laughter (M = .47, SD = .24), ΔM = .05 (95% CI [0.03, 0.06], p < .001). Note that these results are in the same direction as the response biases (i.e., a tendency to choose genuine in case of Dutch laughter and polite in case of Japanese laughter).
There was also a main effect of sex, F(1,495) = 7.36, 95% CI [0.01, 0.07], p < .001, partial η² = .02. Post hoc analysis with a Bonferroni adjustment showed that overall girls (M = .55, SD = .23) performed better than boys (M = .51, SD = .23). There were no interaction effects between Culture and sex or Genuineness and sex.
Adults. There was a main effect of Culture, F(1,263) = 19.15, 95% CI [0.13, 0.17], p < .001, partial η² = .07. Post hoc analysis with a Bonferroni adjustment showed that Dutch
laughter was better recognized (M = .67, SD = .20) than Japanese laughter (M = .59, SD = .21). There was also a main effect of Genuineness, F(1,263) = 8.44, 95% CI [0.01, 0.03], p < .005, partial η² = .03. Post hoc analysis with a Bonferroni adjustment showed that genuine laughter was better recognized (M = .64, SD = .20) than polite laughter (M = .62, SD = .21). However, the difference between genuine and polite laughter (ΔM = .03, 95% CI [0.01, 0.04], p < .001) was only significant for Dutch laughter, adults’ recognition levels were the same for genuine and polite Japanese laughter. There was no main effect of sex, nor were there any interaction effects with sex.
So, the expectation that genuine laughter from both one’s own and a foreign culture would be better recognized compared to the recognition of polite laughter was not fully supported. Children as well as adults recognized genuine Dutch laughter the best, followed by polite Dutch laughter. Adults showed the same level of recognition for genuine and polite Japanese laughter, whereas children were better at recognizing polite Japanese laughter than genuine Japanese laughter.
Figure 2. Children’s mean recognition accuracy as a function of culture (Dutch or Japanese), genuineness (genuine or polite) and age.
0,3 0,35 0,4 0,45 0,5 0,55 0,6 0,65 0,7 0,75 3-6 7 8 9 10 11-12 13-18 L eve l of r ec ogn it ion Age in years
Figure 3. Adults’ mean recognition accuracy as a function of culture (Dutch or Japanese), genuineness (genuine or polite) and age.
Positivity Ratings
To test hypotheses 4a (genuine laughter is perceived as more positive than polite laughter) and 4b (Dutch laughter is perceived as more positive than Japanese laughter), the scores on the third dependent variable “Perceived Positivity” were used. A two-way repeated measures ANOVA was conducted with Culture (Dutch versus Foreign) and Genuineness (Genuine versus Polite) as the within-subject variables and participant’s age and sex as between-subjects variables. Participant’s sex was added as a factor in the analysis to explore whether there were any sex differences. The ANOVA was done for children and adults separately.
Children. There was a strong main effect of Culture, F(1,496) = 143.64, p < .001, 95% CI [0.36, 0.50], partial η² = .23. Post hoc analysis with a Bonferroni adjustment showed that Dutch laughter was rated more positively (M = 4.35, SD = .84) than Japanese laughter (M = 3.92, SD = .89). There was an even stronger main effect of Genuineness, F(1,496) = 625.92, p < .001, 95% CI [1.05, 1.23], partial η² = .56. Post hoc analysis with a Bonferroni adjustment showed that genuine laughter (M = 4.71, SD = .94) was rated as more positive than polite
0,3 0,35 0,4 0,45 0,5 0,55 0,6 0,65 0,7 0,75 19-35 36-40 41-45 46-50 51-75 L eve l of r ec ogn it ion Age in years
laughter (M = 3.57, SD = .90). In addition, an interaction effect was found between
Genuineness and Culture, F(1,496) = 70.94, p < .001, partial η² = .13, due to a much larger difference in perceived positivity (ΔM = 1.42) of genuine and polite Dutch laughter, compared to the difference in perceived positivity of genuine and polite Japanese laughter (ΔM = .87). There was also a main effect of Age, F(1,495) = 12.96, p < .001, partial η² = .03. In addition, an interaction effect was found between Genuineness and Age, F(1,495) = 50.20, p < .001, partial η² = .09. As can be seen in Figure 4, genuine Dutch and genuine Japanese laughter were rated more positively with age, whereas both types of polite laughter were rated less positive with age.
Adults. There was a main effect of Culture, F(1,264) = 45.85, p < .001, 95% CI [0.18, 0.32], partial η² = .15. Post hoc analysis with a Bonferroni adjustment showed that Dutch laughter was rated more positively (M = 4.25, SD = .64) than Japanese laughter (M = 4.00, SD = .72). There was a strong main effect of Genuineness, F(1,264) = 1533.16, p < .001, 95% CI [1.68, 1.86], partial η² = .85. Post hoc analysis with a Bonferroni adjustment showed that genuine laughter (M = 5.00, SD = .68) was rated as more positive than polite laughter (M = 3.24, SD = .75). There was also an interaction effect between Genuineness and Culture, F(1,264) = 93.16, p < .001, partial η² = .26. This effect was again due to a larger difference in perceived positivity (ΔM = 2.12) of genuine and polite Dutch laughter, compared to the difference in perceived positivity of genuine and polite Japanese laughter (ΔM = 1.42). There was no main effect of age (see Figure 4).
These findings offer support for the hypothesis that genuine laughter is perceived as more positive than polite laughter (hypothesis 4a). The results also confirm the expectation that both genuine and polite laughter are judged as positive in valence. No support was found for the expectation that younger children rate polite laughter as less positive than older children; on the contrary, it was the other way around. The hypothesis that laughter from
one’s own culture is perceived as more positive than laughter from a foreign culture is also supported (hypothesis 4b).
Figure 4. Perceived positivity as a function of culture (Dutch or Japanese), genuineness (genuine or polite) and age.
Discussion
The current study set out with three main aims. The first aim was to examine the development of the ability to (a) distinguish between and (b) recognize genuine and polite laughter in children and adults. The second aim was to determine whether exposure to laughter improves the ability to distinguish between types of laughter. The third aim was to investigate whether laugher type influences the perceived positivity of the laughter. The findings will be discussed in this order.
Development of Discriminative Ability with Age
As expected, the youngest children (three- to five-year-olds) in this study were not able to distinguish between genuine and polite laughter. Between five and six years of age, a large increase in discriminative ability occurred. From six years of age, children could distinguish
2,5 3 3,5 4 4,5 5 5,5 6 3-6 7 8 9 10 11-12 13-18 19-35 36-40 41-45 46-50 51-75 P os it ivi ty r at in g Age in years
between genuine and polite laughter. This provides indirect support for the notion that children need first-order Theory of Mind reasoning about other’s belief states in order to be able to distinguish between genuine and polite laughter. This ability emerges around five years of age (Sodian & Kristen, 2009). However, due to the small number of very young children in the sample, these results should be treated with some caution.
Children (range 3-18) showed a general increase in discriminative ability with age; older children were better at distinguishing between genuine and polite laughter than younger children. Adults showed an improvement in discriminative ability until about the age of 40, followed by a stagnation (and eventually decline) from age 40 onwards. This is in line with findings by Scott et al. (in prep), who found that the ability to distinguish between genuine and polite laughter keeps improving until people are in their late thirties.
Role of Social Learning
The increase in discriminative ability in adulthood was only present for Dutch laughter. Adults showed no improvement in discriminative ability for different types of Japanese laughter. This provides support for a social learning hypothesis that would hold that, because people will be frequently exposed to laughter from their own culture, they will improve in their ability to distinguish between different types of laughter from their own culture. Since they will be less exposed to laughter from cultures other than their own (in this case
Japanese), it was expected that there would be less learning for foreign laughter. Additional support for this hypothesis was provided by the finding that participants across the entire age span were better in distinguishing Dutch laughter as compared to Japanese laughter. This is in line with previous studies (Elfenbein et al., 2002; Sauter et al., 2010) that have found an in-group advantage for the recognition of emotional vocalizations from the perceiver’s own cultural group in general.
Recognition of Genuine versus Polite Laughter
This study also looked at specific differences in the ability to recognize genuine versus polite laughter. Beforehand, we expected that, due to the deep biological roots of genuine laughter and the presumably greater influence of culture on polite laughter, genuine laughter from both cultures would be better recognized than polite laughter. This hypothesis was not entirely supported by our findings. Children as well as adults recognized genuine Dutch laughter the best, followed by polite Dutch laughter, which is in line with previous findings in which genuine laughter from one’s own culture was better recognized that polite laughter from one’s culture (Bryant & Aktipis, 2014; Scott et al., in prep). However, genuine Japanese laughter was not better recognized than (Dutch or Japanese) polite laughter. Adults recognized genuine and polite Japanese laughter equally well, whereas children were even better at recognizing polite Japanese laughter than genuine Japanese laughter. It must be noted that this finding may have been influenced by the small response biases that we found (which are in the same direction as these findings). Children’s tendency to choose polite more often when the laughter was Japanese could explain this counterintuitive finding.
Sex Differences
Another unexpected finding of this study was the sex difference in distinguishing between and recognition of genuine and polite laughter in childhood. Girls performed better than boys. This difference was no longer present in the adult sample, meaning that boys did catch up, so it appears to be caused by differential development and not a matter of female sensitivity (a female advantage has been found in some other areas of research, e.g., mind reading abilities in adulthood, Baron-Cohen & Hammer, 1997). Some studies have found a female advantage for theory of mind development (Banerjee, 1997; Bosacki & Astington, 1999), and more
specifically false belief understanding (Cutting & Dunn, 1999; Happé, 1995). However, most studies found no significant advantage for girls or boys (e.g. Holmes, Black, & Miller, 1996; Jenkins & Astington, 1996). Charman, Ruffman, and Clements (2002) have argued that this is because most theory of mind research usually employs samples with less than 50 subjects, which leads to a lack of statistical power to demonstrate such a gender effect. They therefore used two large datasets (N = 375 and N = 1093) and found a weak but significant female advantage for false belief task performance (Charman et al., 2002). They argued that the socialization account provides an explanation for this advantage. A difference in social milieu and social experience (the amount of supportive and emotion talk girls versus boys receive) causes an initial boost in acquiring theory of mind understanding. This advantage is merely temporary and disappears as boys eventually acquire a sufficient quality and quantity of social experience as well (Charman et al., 2002). This explanation fits the findings from the current study well, since our findings show a female advantage in the ability to recognize and
distinguish between genuine and polite laughter in childhood, but this difference disappeared in older children.
Perceived Positivity of Laughter
Finally, we looked at the perceived positivity of the different laughter types. As expected, (a) genuine and polite laughter were both rated as positive in valence, (b) genuine laughter was rated as more positive than polite laughter, which is in accordance with predictions made by several previous theorists (Bachorowski & Owren, 2001; Gervais & Wilson, 2005; Ruch & Ekman, 2001) and the findings of McGettigan et al. (2013), and (c) Dutch laughter was perceived as more positive than Japanese laughter.
Our expectation that older children would rate polite laughter as less negative than younger children was not confirmed. We had based this expectation on findings from studies
on the perception of white lies, in which white lies were rated less negatively with increasing age (Popliger et al., 2011; Xu, Boa, Fu, Talwar, & Lee, 2010). However, white lies are usually examined in a politeness context, in which the prosocial motivation for lying is clear, whereas in our study children heard the polite laughter without any context. It is therefore possible that with increasing age polite laughter was increasingly more negatively perceived because it was only perceived as non-genuine, rather than as a prosocial (i.e., polite) form of behavior. At least in a design in which no further context is provided, it seems to be the case that as children’s ability to distinguish between and recognize genuine and polite laughter improves, the perceived positivity of genuine laughter increases and that of polite laughter decreases. From there on, the perceived positivity of genuine and polite laughter remains unchanged into adulthood.
Notably, the positivity ratings of the four types of laughter are in the same direction as our expectation regarding the recognition of the four laughter types. As discussed above, we expected that genuine laughter from both cultures (with higher recognition performance for genuine Dutch laughter than for genuine Japanese laughter) would be better recognized than Dutch and Japanese polite laughter, which was not the case. However, genuine Dutch laughter was perceived as most positive, followed by genuine Japanese laughter, followed by both types of polite laughter (as can be seen in Figure 4). A possible explanation for these findings could be that people subconsciously correctly recognize each laughter type, despite giving the wrong answer when asked what type of laughter they hear.
Limitations and Future Directions
The current study has some limitations. First of all, participants were not selected at random. The sample consisted of visitors of the science center that were willing to participate. This resulted in a non-equal distribution of participants across different age levels. There were
three age groups that were particularly underrepresented: under the age of seven, between 19 and 35 and above 50. This is unfortunate because all three age groups can provide us with useful knowledge of the development of the ability to distinguish between genuine and polite laughter. Participation of very young children is needed to be able to examine when this ability precisely emerges and how (fast) it develops in early childhood. The group between 19 and 35 is also very relevant and interesting to study, since previous research (Scott et al. in prep) has found that this ability keeps improving until people are in their late thirties. Furthermore, people above 50 years of age could provide interesting information about a potential decrease in this ability after a certain age, possibly due to a decline in general cognitive ability and/or hearing. Therefore, the results regarding these age groups should be taken with some caution, and should ideally be replicated with a larger sample.
A second limitation is on the measurement level. In this study, a forced-choice task was used. Participants were asked whether the laughter samples were genuine or polite (and Dutch or foreign). Providing participants with just two options, one of which is correct, could result in higher recognition levels of both types of laughter than if more than two options had been given. It would be interesting to add more laughter types to the current design, so that it is not a two-alternative forced-choice but for example a four-alternative forced-choice task; previous research has shown that listeners are able to correctly classify various categories of laughter sounds that express different emotions, for example joy, tickling, schadenfreude, and taunting (Szameitat, Alter, Szameitat, Darwin, Wildgruber, Dietrich & Sterr, 2009).
Furthermore, in this study we have measured discrimination and recognition, but not identification. A design with a two-way forced-choice task likely yields different results than a design with an ended task (“What type of laughter do you hear?”). The use of an open-ended task would probably make it harder to correctly recognize and distinguish between genuine and polite laughter, since there are many more laughter types to consider. In addition,
it is possible that the use of such a design would show different results with regard to the developmental pattern of children’s emerging ability to distinguish between genuine and polite laughter; children in particular may have a harder time recognizing polite laughter if they are not told that the laughter is either genuine or polite. As mentioned before, children start to understand that people do not mean everything they express around five years of age (e.g., Josephs, 1994), which is in line with the findings from this study that children’s ability to distinguish between genuine and polite laughter emerges around six years of age. However, it is not until children develop the ability to attribute second-order beliefs, which develops between six and eight years of age (Perner & Wimmer, 1985), that they are able to understand why someone would express something that is not in line with his or her feelings (the
underlying motive) (e.g., Broomfield et al., 2002; Talwar & Crossman, 2011).
This two-phase development has been found in display rule understanding of facial expressions and speech. For example, researchers who studied the use and understanding of white lies in children found that, when asked why they lied (in response to receiving an unwanted gift), older children reported politeness (i.e., not wanting to hurt the gift giver’s feelings) more often as the reason for lying, whereas younger children more often listed self-protective reasons (e.g., so the gift giver would not get angry) (Popliger et al., 2011; Talwar et al., 2007; Xu et al., 2010). It could be that the same two-phase development exists for the recognition of polite laughter, in that the child first understands that a laugh is not genuine, and only later understands the underlying motive, that it is intended to be polite. This would lead to a situation in which the child correctly judges the laughter sample as polite in the current design (because it is recognized as non-genuine), but would not necessarily recognize it as polite in an open-ended task.
We recognize that free-labeling tasks might be too challenging for young children, since they rely heavily on vocabulary. A second issue with an open-ended task is that for
