Consolation has thus far been demonstrated in great apes only. When de Waal and Aureli (1996) set out to apply exactly the same observation methodology as used on chimpanzees to detect consolation in macaques, they failed to find any (reviewed by Watts et al. 2000). This came as a surprise, because reconciliation studies, which employ essentially the same data collection method, have shown reconciliation in species after species. Why, then, would consolation be restricted to apes?
Possibly, one cannot achieve cognitive empathy without a high degree of self-awareness. Targeted help in response to specific, sometimes novel, situations may require a distinction between self and other that allows the other’s situation to be divorced from one’s own while maintaining the emotional link that motivates behavior. In other words, in order to understand that the source of vicarious arousal is not oneself but the other and to understand the causes of the other’s state, one needs a clear distinction between self and other. Based on these assumptions, Gallup (1982) was the first to speculate about a connection between cognitive empathy and mirror self-recognition (MSR). This view is supported both developmentally, by a correlation between the emergence of MSR in young children and their helping tendencies (Bischof-Kohler 1988; Zahn-Waxler et al. 1992), and phylogenetically, by the presence of complex helping and consolation in hominoids (i.e., humans and apes) but not monkeys. Hominoids are also the only primates with MSR.
I have argued before that, apart from consolation behavior, targeted helping reflects cognitive empathy. Targeted helping is defined as altruistic behavior tailored to the specific needs of the other in novel situations, such as the previously described reaction of Kuni to the bird or Binti Jua’s rescue of a boy. These responses require an understanding of the specific predicament of the individual needing help. Given the evidence for targeted helping by dolphins (see above), the recent discovery of MSR in these mammals (Reiss and Marino 2001) supports the proposed connection between increased self-awareness, on the one hand, and cognitive empathy, on the other.
Russian Doll Model
The literature includes accounts of empathy as a cognitive affair, even to the point that apes, let alone other animals, probably lack it (Povinelli 1998; Hauser 2000). This view equates empathy with mental state attribution and ToM. The opposite position has recently been defended in relation to autistic children, however. Contra earlier assumptions that autism reflects a ToM deficit (Baron-Cohen 2000), autism is noticeable well before the age of 4 years at which ToM typically emerges. Williams et al. (2001) argue that the main deficit of autism concerns the socio-affective level, which in turn negatively impacts sophisticated downstream forms of interpersonal perception, such as ToM. Thus, ToM is seen as a derived trait, and the authors urge more attention to its antecedents (a position now also embraced by Baron-Cohen 2003, 2004).
Preston and de Waal (2002a) propose that at the core of the empathic capacity is a relatively simple mechanism that provides an observer (the “subject”) with access to the emotional state of another (the “object”) through the subject’s own neural and bodily representations. When the subject attends to the object’s state, the subject’s neural representations of similar states are automatically activated. The closer and more similar subject and object are, the easier it will be for the subject’s perception to activate motor and autonomic responses that match the object’s (e.g., changes in heart rate, skin conductance, facial expression, body posture). This activation allows the subject to get “under the skin” of the object, sharing its feelings and needs, which embodiment in turn fosters sympathy, compassion, and helping. Preston and de Waal’s (2002a) Perception-Action Mechanism (PAM) fits Damasio’s (1994) somatic marker hypothesis of emotions as well as recent evidence for a link at the cellular level between perception and action (e.g., “mirror neurons,” di Pelligrino et al. 1992).
The idea that perception and action share representations is anything but new: it goes as far back as the first treatise on Einfuhlung, the German concept translated into English as “empathy” (Wispe 1991). When Lipps (1903) spoke of Einfuhlung, which literally means “feeling into,” he speculated about innere Nachahmung (inner mimicry) of another’s feelings along the same lines as proposed by the PAM. Accordingly, empathy is a routine involuntary process, as demonstrated by electromyographic studies of invisible muscle contractions in people’s faces in response to pictures of human facial expressions. These reactions are fully automated and occur even when people are unaware of what they saw (Dimberg et al. 2000). Accounts of empathy as a higher cognitive process neglect these gut-level reactions, which are far too rapid to be under conscious control.
Perception-action mechanisms are well known for motor perception (Prinz and Hommel 2002), causing researchers to assume similar processes to underlie emotion perception (Gallese 2001; Wolpert et al. 2001). Data suggest that both observing and experiencing emotions involves shared physiological substrates: seeing another’s disgust or pain is very much like being disgusted or in pain (Adolphs et al. 1997, 2000; Wicker et al. 2003). Also, affective communication creates similar physiological states in subject and object (Dimberg 1982, 1990; Levenson and Reuf 1992). In short, human physiological and neural activity does not take place on an island, but is intimately connected with and affected by fellow human beings. Recent investigations of the neural basis of empathy lend strong support to the PAM (Carr et al. 2003; Singer et al. 2004; de Gelder et al. 2004).
Figure 4 According to the Russian Doll Model, empathy covers all processes leading to related emotional states in subject and object. At its core is a simple, automatic Perception-Action Mechanism (PAM), which results in immediate, often unconscious state matching between individuals. Higher levels of empathy that build on this hardwired basis include cognitive empathy (i.e., understanding the reasons for the other’s emotions) and mental state attribution (i.e., fully adopting the other’s perspective). The Russian Doll Model proposes that outer layers require inner ones. After de Waal (2003).
How simple forms of empathy relate to more complex ones has been depicted as a Russian doll by de Waal (2003). Accordingly, empathy covers all forms of one individual’s emotional state affecting another’s, with basic mechanisms at its core and more advanced mechanisms and cognitive abilities as its outer layers (figure 4). Autism may be reflected in deficient outer layers of the Russian doll, but such deficiencies invariably go back to deficient inner layers.
This is not to say that higher cognitive levels of empathy are irrelevant, but they are built on top of this firm, hardwired basis without which we would be at a loss about what moves others. Surely, not all empathy is reducible to emotional contagion, but it never gets around it. At the core of the Russian doll, we find a PAM-induced emotional state that corresponds with the object’s state. In a second layer, cognitive empathy implies appraisal of another’s predicament or situation (cf. de Waal 1996). The subject not only responds to the signals emitted by the object, but seeks to understand the reasons for these signals, looking for clues in the other’s behavior and situation. Cognitive empathy makes it possible to furnish targeted help that takes the specific needs of the other into account (figure 5). These responses go well beyond emotional contagion, yet they would be hard to explain without the motivation provided by the emotional component. Without it, we would be as disconnected as Mr. Spock in Star Trek, constantly wondering why others feel what they say they feel.
Whereas monkeys (and many other social mammals) clearly seem to possess emotional contagion and a limited degree of targeted helping, the latter phenomenon is not nearly as robust as in the great apes. For example, at Jigokudani Monkey Park, in Japan, first-time mother macaques are kept out of the hot springs by park wardens because of the experience that these females will accidentally drown their infants. They fail to pay attention to them when submerging themselves in the ponds. This is something monkey mothers apparently have to learn with time, showing that they do not automatically take their offspring’s perspective. De Waal (1996
) ascribed their behavioral change to “learned adjustment,” setting it apart from cognitive empathy, which is more typical of apes and humans. Ape mothers respond immediately and appropriately to the specific needs of their offspring. They are, for example, very careful to keep them away from water, rushing over to pull them away as soon as they get too close.
Figure 5 Cognitive empathy (i.e., empathy combined with appraisal of the other’s situation) allows for aid tailored to the other’s needs. In this case, a mother chimpanzee reaches out to help her son out of a tree after he has screamed and begged (see hand gesture). Targeted helping may require a distinction between self and other, an ability also thought to underlie mirror self-recognition, as found in humans, apes, and dolphins. Photograph by the author.
In conclusion, empathy is not an all-or-nothing phenomenon: it covers a wide range of emotional linkage patterns, from the very simple and automatic to the highly sophisticated. It seems logical to first try to understand the basic forms of empathy, which are widespread indeed, before addressing the variations that cognitive evolution has constructed on top of this foundation.
RECIPROCITY AND FAIRNESS
Chimpanzees and capuchin monkeys—the two species I work with most—are special, as they are among the very few primates that share food outside the mother-offspring context (Feistner and McGrew 1989). The capuchin is a small primate, easy to work with, as opposed to the chimpanzee, which is many times stronger than we are. Members of both species are interested in each other’s food and will share food on occasion—sometimes even hand over a piece to another. Most sharing, however, is passive, where one individual will reach for food owned by another, who will let go. But even passive sharing is special when compared to most animals, for which a similar situation would result in a fight or assertion by the dominant, without any sharing at all.
Chimpanzee Gratitude
We studied sequences involving food sharing to see how a beneficial act by individual A toward B would affect B’s behavior toward A. The prediction was that B would show beneficial behavior toward A in return. The problem with food sharing is, however, that after a group-wide feeding session as used in our experiments, the motivation to share changes (the animals are more sated). Hence, food sharing cannot be the only variable measured. A second social service unaffected by food consumption was included. For this, grooming between individuals prior to food sharing was used. The frequency and duration of hundreds of spontaneous grooming bouts among our chimpanzees were measured in the morning. Within half an hour after the end of these observations, starting around noon, the apes were given two tightly bound bundles of leaves and branches. Nearly 7,000 interactions over food were carefully recorded by observers and entered into a computer according to strict definitions described by de Waal (1989a). The resulting database on spontaneous services exceeds that for any other nonhuman primate.
It was found that adults were more likely to share food with individuals who had groomed them earlier. In other words, if A had groomed B in the morning, B was more likely than usual to share food with A later in the day. This result, however, could be explained in two ways. The first is the “good mood” hypothesis according to which individuals who have received grooming are in a benevolent mood, leading them to share indiscriminately with all individuals. The second explanation is the direct-exchange hypothesis, in which the individual who has been groomed responds by sharing food specifically with the groomer. The data indicated that the sharing increase was specific to the previous groomer. In other words, chimpanzees appeared to remember others who had just performed a service (grooming) and respond to those individuals by sharing more with them. Also, aggressive protests by food possessors to approaching individuals were directed more at those who had not groomed them than at previous grooming partners. This is compelling evidence for partner-specific reciprocal exchange (de Waal 1997b).
Of all existing examples of reciprocal altruism in nonhuman animals, the exchange of food for grooming in chimpanzees appears to be the most cognitively advanced. Our data strongly suggest a memory-based mechanism. A significant time delay existed between favors given and received (from half an hour to two hours); hence the favor was acted upon well after the previous interaction. Apart from memory of past events, we need to postulate that the memory of a received service, such as grooming, triggered a positive attitude toward the individual who offered this service, a psychological mechanism known in humans as “gratitude.” Gratitude within the context of reciprocal exchange was predicted by Trivers (1971), and has been discussed by Bonnie and de Waal (2004). It is classified by Westermarck (1912 [1908]) as one of the “retributive kindly emotions” deemed essential for human morality.
Monkey Fairness
During the evolution of cooperation it may have become critical for actors to compare their own efforts and payoffs with those of others. Negative reactions may ensue in case of violated expectations. A recent theory proposes that aversion to inequity can explain human cooperation within the bounds of the rational choice model (Fehr and Schmidt 1999). Similarly, cooperative nonhuman species seem guided by a set of expectations about the outcome of cooperation and access to resources. De Waal (1996: 95) proposed a sense of social regularity, defined as “A set of expectations about the way in which oneself (or others) should be treated and how resources should be divided. Whenever reality deviates from these expectations to one’s (or the other’s) disadvantage, a negative reaction ensues, most commonly protest by subordinate individuals and punishment by dominant individuals.”
The sense of how others should or should not behave is essentially egocentric, although the interests of individuals close to the actor, especially kin, may be taken into account (hence the parenthetical inclusion of others). Note that the expectations have not been specified: they tend to be species-typical. For example, a rhesus monkey expects no share of a dominant’s food, as it lives in a despotically hierarchical society, but a chimpanzee definitely does, hence the begging, whining, and temper tantrums if no share is forthcoming. I consider expectations the most important unstudied topic in animal behavior, which is all the more lamentable as it is the one issue that will bring animal behavior closest to the “ought” of behavior that we recognize so clearly in the moral domain.
To explore the expectations held by capuchin monkeys, we made use of their ability to judge and respond to value. We knew from previous studies that capuchins easily learn to assign value to tokens. Furthermore they can use these assigned values to complete a simple barter. This allowed a test to elucidate inequity aversion by measuring the reactions of subjects to a partner receiving a superior reward for the same tokens.
We paired each monkey with a group mate and watched their reactions when their partners got a better reward for doing the same bartering task. This consisted of an exchange in which the experimenter gave the subject a token that could immediately be handed back for a reward (figure 6). Each session consisted of twenty-five exchanges by each individual, and the subject always saw the partner’s exchange immediately before its own. Food rewards varied from lower value rewards (e.g., a cucumber piece), which they are usually happy to work for, to higher value rewards (e.g., a grape), which were preferred by all individuals tested. All subjects were subjected to (a) an Equity Test (ET), in which subject and partner did the same work for the same low-value food, (b) an Inequity Test (IT), in which the partner received a superior reward (grape) for the same effort, (c) an Effort Control Test (EC), designed to elucidate the role of effort, in which the partner received the higher value grape for free, and (d) a Food Control Test (FC), designed to elucidate the effect of the presence of the reward on subject behavior, in which grapes were visible but not given to another capuchin.
Figure 6 A capuchin monkey in the test chamber returns a token to the experimenter with her right hand while steadying the human hand with her left hand. Her partner looks on. Drawing by Gwen Bragg and Frans de Waal after a video still.
Figu
re 7 Mean percentage ± standard error of the mean of failures to exchange for females across the four test types. Black bars represent the proportion of nonexchanges due to refusals to accept the reward; white bars represent nonexchanges due to refusals to return the token. ET = Equity Test, IT = Inequity Test, EC = Effort Control, FC = Food Control. The Y-axis shows the percentage of nonexchanges.
Individuals who received lower value rewards showed both passive negative reactions (e.g., refusing to exchange the token, ignoring the reward) and active negative reactions (e.g., throwing out the token or the reward). Compared to tests in which both received identical rewards, the capuchins were far less willing to complete the exchange or accept the reward if their partner received a better deal (figure 7; Brosnan and de Waal 2003). Capuchins refused to participate even more frequently if their partner did not have to work (exchange) to get the better reward but was handed it for “free.” Of course, there is always the possibility that subjects were just reacting to the presence of the higher value food and that what the partner received (free or not) did not affect their reaction. However, in the Food Control Test, in which the higher value reward was visible but not given to another monkey, the reaction to the presence of this high-value food decreased significantly over the course of testing, which is a change in the opposite direction from that seen when the high-value reward went to an actual partner. Clearly our subjects discriminate between higher value food being consumed by a conspecific and such food being merely visible, intensifying their rejections only to the former (Brosnan and de Waal 2003).
Primates and Philosophers_How Morality Evolved Page 5
