Jump to ContentJump to Main Navigation
Altruism in Humans$

C. Daniel Batson

Print publication date: 2010

Print ISBN-13: 9780195341065

Published to Oxford Scholarship Online: May 2011

DOI: 10.1093/acprof:oso/9780195341065.001.0001

Show Summary Details
Page of

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (www.oxfordscholarship.com). (c) Copyright Oxford University Press, 2018. All Rights Reserved. Under the terms of the licence agreement, an individual user may print out a PDF of a single chapter of a monograph in OSO for personal use (for details see www.oxfordscholarship.com/page/privacy-policy). Subscriber: null; date: 16 January 2019

Antecedents of Empathic Concern

Antecedents of Empathic Concern

(p.33) 2 Antecedents of Empathic Concern
Altruism in Humans

C. Daniel Batson

Oxford University Press

Abstract and Keywords

In everyday life, there seem to be two antecedents of empathic concern: (a) perceiving the other as in need and (b) valuing the other’s welfare. This chapter considers each of these antecedents, as well as other possible ones—perceived innocence, similarity, and perspective taking. Cognitive abilities required to perceive need are specified, raising the possibility that only humans have the capacity to experience empathic concern. It is suggested that valuing another’s welfare naturally leads to perspective taking, allowing the latter to serve as a proxy for the former in laboratory research. Valuing of the other’s welfare is linked to human parental nurturance, which is emotion-based and goal-directed. Neurochemistry and neurophysiology of parental care and empathic concern are considered. Individual differences, including gender differences, are viewed as moderators rather than antecedents of empathic concern.

Keywords:   empathic concern, individual differences, innocence, need, perspective taking, parental nurturance, similarity, valuing another’s welfare

If empathic concern produces altruistic motivation, as the empathy-altruism hypothesis claims, then what produces empathic concern? Think back to your experience at lunch described in Chapter 1. What caused you to feel so sorry for your friend? First, she had just lost her job, and was hurt and scared. Second, she was a close friend; you cared what happened to her and how she felt. More generally, in everyday life two conditions seem necessary to feel empathic concern: (a) perceiving the other as in need and (b) valuing the other’s welfare. The relationship of these antecedents is specified in Figure 2.1 with a multiplication sign, indicating that some level of each antecedent is necessary and, beyond this threshold level, the magnitude of empathic concern is a product of the strength of each. Hence, the shared causal path to empathic concern. The relative weighting of the two antecedents, and whether the effect of each is linear or—more likely—negatively accelerated as it approaches some asymptote, is not specified.

Perceiving the Other as in Need

In Chapter 1, I spoke of perceiving another as in need without saying exactly what was meant. Now it is time to be more precise. Perceiving need involves perceiving a negative discrepancy between the other’s current state and what is desirable for the other on one or more dimensions of well-being. Dimensions of well-being include the absence of physical pain, negative affect, anxiety, stress, danger, and disease, as well as the presence of physical pleasure, positive affect, satisfaction, and security.

The negative discrepancy in well-being that is at issue is for the person in need, not for the person feeling empathic concern. But the perception at issue is by the person feeling empathy, not the person in need. There are times when people perceive themselves to be in need, yet others do not. These others will not experience empathic concern—unless they consider the false perception of need itself to be a need. Alternatively, there are times when people do not perceive themselves to be in need, yet others do. These others may well feel empathic concern.

Perception of need seems to be a threshold function of two situational factors: First, the discrepancy (real or apparent) between what is and what is desirable must be noticed (p.34)

                      Antecedents of Empathic Concern

Figure 2.1 Antecedents of Empathic Concern

(Clark & Word, 1972, 1974; Latané & Darley, 1970). Second, attention must be focused on the person in need, not on the self or some other aspect of the situation (Aderman & Berkowitz, 1983; Gibbons & Wicklund, 1982; Mathews & Canon, 1975; Milgram, 1970; Weiner, 1976; Wicklund, 1975). Both of these conditions must simultaneously be satisfied to perceive another’s need.

Satisfying these conditions does not, however, guarantee perception of need. Additional cognitive and situational factors—such as being led by the reactions of other bystanders to misinterpret the situation (“No one else seems upset, so I guess that scream was not a cry of distress, only play.”)—may lead a perceiver to minimize or even deny an apparent need (Latané & Darley, 1970). Additional factors may also facilitate perception of need in ambiguous situations. One such factor, described in Chapter 1, is imagining yourself in the other’s shoes. It may lead you to recognize that, were you in the other’s situation, you would be experiencing need.

Given that need is perceived, it can vary in magnitude. The magnitude appears to be a function of three factors: (a) the number of dimensions of well-being on which discrepancies are perceived, (b) the size of each discrepancy, and (c) the perceived importance of each of these dimensions for the overall well-being of the person in need.

Is Perceived Innocence Necessary for Empathic Concern?

Weiner (1980) and Nussbaum (2001) both contend that in addition to perceiving a person as in need we must also perceive him or her to be free of responsibility for causing the need before we can feel empathic concern (sympathy, pity, compassion). For them, perception of need and perception of innocence are two separate antecedents. Yet, as Nussbaum notes, we readily feel compassion for those for whom we especially care, even when they bring suffering on themselves. Nussbaum attempts to handle such cases by suggesting that we see these others, such as our children, as being at a stage of life where they cannot be held responsible. But she fails to consider those cases in which we feel (p.35) compassion for adults for whom we especially care—close friends, lovers, spouses, siblings, parents—even when they bring suffering on themselves. This omission is, I think, a problem. (In response to John Deigh, 2004, Nussbaum, 2004, modified her position to allow for compassion without consideration of blame, but she did not deal with cases in which we feel compassion for loved ones in spite of knowing they caused their own need.)

I propose a different analysis. Rather than two separate antecedents, I believe that perceived responsibility contributes to our perception of need. In general, we may feel that people who bring suffering on themselves get what they deserve. If so, and if we also believe that people should get what they deserve (Lerner, 1970), then there is no discrepancy between our perception of their current state and the state we deem desirable for them. We perceive no need. But deservedness is not the only dimension of well-being on which we assess the state of those for whom we care. Discrepancies on these other dimensions may lead to a perception of need—and to empathic concern—even when there is no perceived discrepancy on deservedness.


One form of need that should get special mention is vulnerability. Even when there is no specific immediate discrepancy between what is and what is desirable, another may be perceived to be vulnerable to such discrepancies, which is itself a form of need.

Perception of vulnerability is especially likely if the other is viewed as comparatively defenseless and unaware of danger. Think, for example, of your reaction to seeing a young child happily running across a playground, or to seeing this child safely asleep in bed. Think of your reaction to seeing a puppy in similar situations. No immediate need is apparent. Still, the young child’s—or puppy’s—vulnerability is apt to trigger empathic feelings of tenderness, warmth, and softheartedness. Vulnerability alone does not, however, trigger feelings of sympathy or pity. These latter empathic feelings seem to require the presence of an immediate need (Dijker, 2001; Lishner, 2003).

Evidence That Empathic Concern Requires Perception of Need

Not wanting to take anything for granted, several psychologists conducted research in the 1960s to demonstrate that empathic concern requires perception of need. Berger (1962) had people observe a target person perform a task. He led these people to believe that at the onset of a visual signal the target either received electric shocks (electric-shock condition) or did not (no-shock condition). Further, the target either jerked his arm at the visual signal (movement condition) or did not (no-movement condition). All research participants were told that they themselves would not be shocked during the experiment.

Berger reasoned, first, that both a painful stimulus in the environment (shock) and a distress response (movement) were necessary for an observer to infer that the target was experiencing pain (i.e., need). He reasoned, second, that if participants in his experiment (p.36) were feeling empathic concern for the target, as opposed to feeling fear or anxiety about the shock itself, then they should display a physiological reaction to watching the target only when they inferred that he was experiencing pain. Accordingly, Berger predicted that only participants in the shock/movement condition would display increased physiological arousal because only they would make this inference. For participants in each of the other three conditions, some information necessary to infer pain was missing.

Results followed the predicted pattern. Consistent with the assumption that people can experience empathic concern when attending to another perceived to be in need, participants in the shock/movement condition were more physiologically aroused while observing the target than were participants in the other three conditions. (Physiological arousal was assessed by electrodermal skin conductance.) Berger concluded that empathic arousal occurs in response to perceived need. Subsequent research also supported this conclusion (Bandura & Rosenthal, 1966; Craig & Lowery, 1969; Craig & Wood, 1969). Later, Hygge (1976) tested and found support for the idea that the evoked physiological arousal (again assessed by skin conductance) is a response to what the target is believed to be experiencing and not a response to what participants believe they would themselves experience in the target’s situation (also see Lamm et al., 2010).

It is important to note that results of these studies do not fit an automatic motor-mimicry or neural response matching view of empathic emotion, such as Preston and de Waal’s (2002b) perception-action model (PAM). Physiological responses in these studies were not automatic reactions to what was visually perceived. Instead, the responses were clearly contingent on cognitive interpretation of the target’s experience. They were responses to perception of need.

The research by Berger (1962), Hygge (1976), and others just mentioned demonstrates that people respond physiologically to perceiving another as in need. Stotland (1969) reported a series of experiments supporting two further ideas: (a) this response reflects other-oriented empathic concern, and (b) perspective taking can increase empathic concern. Stotland had research participants watch a male target undergo what they believed was a painful diathermy treatment. Participants instructed to imagine how the target felt (imagine-him condition) or to imagine how they would feel in the target’s place (imagine-self condition) showed more physiological arousal (assessed by vasoconstriction and palmar sweat) and reported feeling more emotion than participants instructed to watch the target’s movements (observe condition). These participants also showed more physiological arousal and reported feeling more emotion than participants given the imagine instructions but led to believe that the diathermy treatment was not painful.

Consistent with the distinction made in Chapter 1 between the imagine-other and imagine-self perspectives, emotional effects of the two forms of perspective taking were not the same, either on physiological or self-report measures. Among participants observing the target undergo what they thought was a painful diathermy experience, those given the imagine-him instructions showed more vasoconstriction, which Stotland (1969) interpreted as evidence that they “were reacting to the feelings they perceived the model as having at a given moment” (p. 296). Those given the imagine-self instructions showed more palmar sweat and reported feeling more tension and nervousness, which (p.37) Stotland interpreted as evidence that their emotional reactions were more self-oriented and “not quite so tied to the experience of the model” (p. 297). Thus, the imagine-him instructions seemed to evoke relatively pure empathic feelings, whereas the imagine-self instructions seemed to evoke more personal distress (for parallel findings, see Batson, Early, & Salvarani, 1997).

Who Can Perceive Another as in Need?

Cognitive Abilities Required

Surprisingly perhaps, perception of another as in need may be a uniquely human skill. If so, and if this perception is a necessary antecedent of empathic concern, then empathic concern and empathy-induced altruism must be uniquely humans also. Consider the cognitive abilities necessary to perceive another as in need. First, one must recognize the other as an animate being who is not only qualitatively different from physical objects but also distinct from other animate beings, including oneself. Apparently, this recognition occurs in the normal child’s first year of life (Hoffman, 1975, 2000). It also occurs early in the normal development of non-human primates—and probably in the development of other higher mammals as well (Tomasello, 1999).

Second, it is necessary to recognize that the other has values, goals, and feelings. Tomasello (1999) speaks of this as understanding that the other is an intentional agent, not merely an animate being. Hoffman (1987, 2000) speaks of awareness that others have feelings and internal states. Povinelli (Povinelli & Bering, 2002; Povinelli et al., 2000) speaks of second-order mental states, which involve inferring intentions. Tomasello (1999) presents evidence that this ability emerges in normal children at around 9–12 months; Hoffman (2000) puts it a bit later, at around 18–24 months (also see Bretherton, McNew, & Beeghly-Smith, 1981; Dunn & Kendrick, 1982; Kagan, 2000; Meltzoff, 1995; Thompson, 1987). Within this age range, infants first begin to recognize that they have goals, intentions, desires, and feelings. Soon thereafter—perhaps because of a uniquely human adaptation that allows them to understand other persons as beings “like me yet distinct from me” (Tomasello, 1999)—infants begin to recognize that others also have goals, intentions, desires, and feelings. With this recognition, infants see others not simply as acting but as acting with purpose, circumventing barriers and using alternative behavioral routes to reach desired goals. (The recognition of desires and feelings in others may be related to the development of Von Economo neurons in the anterior insula and anterior cingulate cortex—see Allman, Watson, Tetreault, & Hakeem, 2005; Craig, 2005.) Often, infants initially extend this perception too far, applying it not only to people but also to toys and machines. Over time, experience hones the perception.

Who Has These Abilities?

Both Tomasello (1999) and Povinelli et al. (2000) concluded based on the existing evidence at the time that the ability to see others as sentient, intentional agents was limited to humans (and possibly some human-reared primates). Other primates have the (p.38) ability to recognize and generalize from contingent stimulus-response relations and to predict the behavior of others (if I kick her, she is likely to kick back), but even chimpanzees seemed to lack the ability to infer the internal mediating processes—feelings, desires, goals, and intentions—that explain why a given stimulus-response sequence occurs (she doesn’t like to be kicked). Humans have both.

Tomasello (1999) further suggested that the ability to perceive others as intentional agents is the basis for cultural learning and cultural evolution. It is what allows humans (a) to learn ways of dealing with the world through observing others, (b) to learn goal-directed strategies from others, (c) to be able to creatively modify these strategies, and (d) to be able to recognize and retain improvements. If all this is true, then the ability to recognize others as intentional agents is one truly giant step for mankind. Moreover, if only humans have this capacity, then only humans, and no other species, are capable of experiencing empathic concern and the altruistic motivation it produces. Recognition of the other as a sentient, intentional agent with values, goals, desires, and feelings is necessary to perceive discrepancies on dimensions of well-being, because without this recognition, issues of the other’s well-being do not arise. There can be no perception of need and, hence, no empathic concern.

But one must be cautious about claiming that a given ability is uniquely human. Such claims have often proved false. Indeed, Tomasello has more recently withdrawn his claim that chimpanzees lack the ability to see others as sentient, intentional agents (Call & Tomasello, 2008; Tomasello, Call, & Hare, 2003), although Povinelli has not (Penn, Holyoak, & Povinelli, 2008; Povinelli & Vonk, 2003). Tomasello now argues that evidence from experiments modeled more closely on situations that chimpanzees routinely encounter in their natural environment, such as competition for food, do provide evidence that at least human-reared chimpanzees understand others as having goals and intentions. Call and Tomasello (2008) claim that the new experimental data cannot be explained by arguing that chimpanzees rely on behavioral rules, even circumstance-contingent behavioral rules.

However, it seems that these data can be explained if the chimpanzees in the new experiments are applying learned knowledge of abstract contingent action-consequence relations (“If I reach toward the food in front of him, he’ll grab and eat it, but not if his head is turned.” “If either Action A or Action B would produce Consequence C, and she chooses A, then it may be the better means to produce C.”). Use of such knowledge occurs in a wide range of mammalian species and does not require inferences about another’s goals or intentions, only probabilistic knowledge of contingent action-consequence relations. Thus, even the more recent experimental data have not provided clear evidence that chimpanzees infer intentions. The question of whether only humans see others as intentional agents remains unresolved (Call & Tomasello, 2008; Penn et al., 2008).

Frans de Waal (1996) provided a nice illustration of how hard it is to resist the assumption that when other animals display behavior similar to what we might display in a given situation, their behavior must be the product of equally complex psychological processes. Most dog owners and lovers, myself included, are quite familiar with canine guilt. Mango, a Siberian husky, displayed such guilt after persisting in shredding newspapers, (p.39) magazines, and books, despite scolding and punishment. Mango’s owner thought Mango knew the shredding was wrong and was acting out of spite for having been left alone. Peter Vollmer (1977), an animal behavior consultant, used a simple demonstration to show that although Mango behaved as if she were feeling spite and guilt, her behavior was a product of a learned contingent action-consequence relation. With Mango out of the house, Vollmer had her owner shred some newspapers. Mango was then let back in the house, and her owner left for fifteen minutes. On the owner’s return, Mango acted as guilty as when she shredded things herself. In de Waal’s words: “The only thing she seemed to understand was: Evidence + Owner = Trouble” (1996, pp. 107–108). We are easily misled about the psychological processes underlying a behavioral sequence when it is taken out of the context of situational cues and learning history. (Sapolsky, 2010, provides other nice examples.)

For present purposes, the ongoing debate about whether other species infer intentions or simply use knowledge of contingent action-consequence relations to behave as if they infer intentions highlights the point that in order to have evidence of empathic concern and empathy-induced altruism it is not enough to observe a response by one organism to another’s distress cries, even a response intended to stop those cries. One must have evidence that the other’s plight is more than a noxious stimulus or a conditioned cue for danger, as it appears to have been in the research claiming evidence of “altruism” in mice, rats, and monkeys described in Chapter 1 (also see Langford, Crager, Shehzad, Smith, Sotocinal, Levenstadt, Chanda, Levitin, & Mogil, 2006) as well as in research showing that one chimpanzee will console the loser of a fight (de Waal, 1996, 2008; Romero, Castellanos, & de Waal, 2010) or will help another get food (Warneken, Hare, Melis, Hanus, & Tomasello, 2007, Experiment 3). There must be perception of need, other-oriented emotion, and a goal-directed desire to remove the need.

What About Primates, Elephants, Dolphins, and Dogs?

Some primatologists believe they see evidence of sensitivity to the feelings and intentions of others—and evidence of empathy and altruism—among at least some primate species other than humans (de Waal, 1996, 2008, 2009; de Waal, Leimgruber, & Greenberg, 2008; Goodall, 1990; Kohler, 1927; Povinelli, 1993; Preston & de Waal, 2002a; for contrary evidence see Brosnan, Silk, Henrich, Mareno, Lambeth, & Schapiro, 2009; Jensen, Hare, Call, & Tomasello, 2006; Povinelli & Bering, 2002; Povinelli et al., 2000; Silk, Brosnan, Vonk, Henrich, Povinelli, Richardson, Lambeth, Mascaro, & Schapiro, 2005; Vonk, Brosnan, Silk, Henrich, Richardson, Lambeth, Schapiro, & Povinelli, 2008; Warneken & Tomasello, 2006, but also see Warneken et al., 2007). Other experts make similar claims for elephants (Bates et al., 2008; de Waal, 1996; Moss, 2000; Poole, 1997), or for dolphins and whales (Caldwell & Caldwell, 1966; Connor & Norris, 1982; Hindley, 1985; McIntyre, 1974; Trivers, 1985, pp. 382–386, but also see Wilson, 1975, pp. 474–475). Darwin (1871, p. 104), who was a great lover of dogs, thought that he could see such sensitivity when a dog gently licks the face of an injured master or an ailing canine friend (but see Macpherson & Roberts, 2006).

(p.40) However, careful examination of existing evidence reveals that even the most touching and tantalizing examples of nonhuman response to another in need fail to show unequivocal awareness that the other is in need—let alone empathic emotion and altruistic motivation. Consider the famous 1996 case of Binti Jua, the 8-year-old female gorilla at the Brookfield Zoo outside Chicago who rescued and gently held an unconscious 3-year-old boy after he fell into the primate enclosure. Binti Jua eventually turned the child over to the zoo staff unharmed. As Joan Silk (2009) explains:

Some have cited this incident as evidence for empathy and sympathy in apes, arguing that Binti Jua was motivated by compassion and concern for the welfare of the child (Preston & de Waal, 2002). However, other facts need to be considered. Binti Jua was hand-reared by humans, after being rejected by her own mother. Concerned that Binti Jua might become a neglectful mother herself, the zoo staff used operant training methods to guide the development of appropriate maternal skills. One of the things that she was trained to do was to retrieve a doll-like object and bring it to the front of the enclosure, where the zoo personnel could inspect it. (pp. 275–276)

Did Binti Jua perceive the child’s need and act to meet it, or was her response generalized from the training she received prior to the birth of her own baby? I do not think we know. As with Mango, a behavioral episode taken out of the context of situational cues and learning history can easily be misread.

There are a number of non-human examples that, without knowing more about context, seem to satisfy the conditions for perceiving need (i.e., for understanding the other is a sentient, intentional agent with desires and feelings), providing what de Waal (2008) has called “targeted helping.” One example offered by de Waal (1996, p. 83) is the intriguing observation by Otto Adang of two chimpanzees, Krom and Jakie, at the Arnhem Zoo. Krom, an elderly female, spent over ten minutes pulling and pushing on a rubber tire that held some water. The tire was hanging on a horizontal log extending from a climbing frame. Unfortunately, there were a half-dozen heavy tires hanging in front of the tire with the water, and Krom made no progress in getting it off the log. Jakie, a 7-year-old male whom Krom had taken care of as a juvenile, watched her struggle unsuccessfully with the tire and finally give up. When she walked away, Jakie went over, pushed the tires off the log one by one until he was able to remove the tire with the water. He then carried it straight to Krom, who began scooping the water out with her hand and drinking. Jakie’s behavior seems hard to explain without assuming that he perceived Krom’s need (i.e., he recognized what she wanted but did not have) and acted to meet that need.

A second example suggests possible anticipation of need on the part of an old, experienced male bonobo, Kakowet. The moat around the bonobo enclosure at the San Diego Zoo was routinely drained for cleaning. One day, unnoticed by the keepers, several young bonobos had entered the dry moat and were unable to get out. When the keepers went to open the valve to refill the moat, Kakowet came to the window screaming and waving his arms, alerting the keepers and preventing possible disaster (de Waal, 2006, p. 71).

These examples are certainly intriguing and suggestive, but given our lack of knowledge of context, it is probably unwise to place the heavy weight of a claim that chimpanzees, bonobos, or other species satisfy the conditions for perceiving need on (p.41) these two examples—or on the other examples that might be cited, such as the efforts of elephants to help a wounded or sick member of the herd to rise (Moss, 2000; Poole, 1997) or of dolphins and whales to protect or rescue pod-mates and even humans (Caldwell & Caldwell, 1966; Connor & Norris, 1982). Nor is the experimental evidence to date yet clear enough to warrant such a claim. At least for now, it seems wise to suspend judgment on this important possibility. Fortunately, we need not settle the issue of non-human perception of need to proceed with our analysis of altruism in humans. There is little doubt that by two years of age, most normal human children are capable of perceiving need. They understand that others are sentient and goal directed. Failure to understand this is considered a key feature of autism (Allman et al., 2005; Gillberg, 1992; Klin, 2000; Tomasello, 1999).

Valuing the Other’s Welfare

The two abilities that Tomasello identified—(a) recognize others as distinct, animate beings and (b) recognize them as sentient, intentional agents—enable one to perceive another as in need (as deficient on one or more dimensions of well-being). But to feel empathic concern, more is required; one also needs to care about whether the other is in need and about how this need affects the other’s life. Apparently, in normal humans the capacity to place value on another’s welfare emerges somewhere between one and three years of age (Hoffman, 1975, 2000; Rheingold, 1982; Thompson, 1987; Zahn-Waxler, Radke-Yarrow, & King, 1979; Zahn-Waxler, Radke-Yarrow, Wagner, & Chapman, 1992). When it fails to develop, we speak of psychopathy or sociopathy (Anderson, Bechara, Damasio, Tranel, & Damasio, 1999; Damasio, 1994; also see Blair, 2004, 2007).

Why should one person value another’s welfare, especially someone who is not kin? Is not such valuing a violation of principles of natural selection? Is not interest exclusively in one’s own welfare—in looking out for Number One—a central tenet of the theory of rational choice? We need to consider in some detail what might lead one person to value another’s welfare.

One often hears lip service paid to valuing all human life or the welfare of all humanity (Monroe, 1996). Most of us, however, place different value on the welfare of different others. We value the welfare of some quite highly. We value the welfare of some very little, if at all. We may even place a negative value on the welfare of some, such as a rival.

If we place no value on the welfare of a person perceived to be in need, then we are not likely to think about how this person is affected by the need, except perhaps as a means to control his or her behavior. The perceived need provides no basis for feeling empathic concern—or any other emotion. We understand what the other needs but do not care. This might be called a dispassionate or objective orientation to the other.

If we place negative value on a person’s welfare, which we may if we dislike or are in competition with the person, then perceiving him or her in need will produce emotions quite different from the other-oriented, congruent emotions I am calling empathic concern. We are apt to feel pleasure at the person’s suffering, or even the malicious glee (p.42) called schadenfreude (James, 1890; Lanzetta & Englis, 1989; Singer, Seymour, O’Doherty, Stephan, Dolan, & Frith, 2006; Zillmann & Cantor, 1977). In this case, although we may be well aware of the other person’s desires and feelings about his or her situation, we do not adopt an other-oriented value assessment of events. Instead, our assessment is antithetical to this person’s welfare. This might be called a hostile orientation.

If we positively value a person’s welfare, then we are likely to think about how this person is affected by the events in his or her life, and to adopt an other-oriented value assessment of these events. By an other-oriented value assessment, I mean one that is congruent with the perceived welfare (well-being) of the other. Positive value is placed on events that we think will bring the person pleasure, joy, satisfaction, safety, or relief; negative value is placed on events that we think will bring the person pain, sorrow, discontent, danger, or disappointment. Such valuing not only produces a lively response to events that affect this person’s welfare, much as we might respond to events that affect our own welfare, but it also produces vigilance. It leads us naturally to adopt his or her perspective, imagining how this person thinks and feels about events. His or her welfare becomes part of our own value structure. This might be called a sympathetic orientation.

Why Not Similarity Instead of Valuing?

Not everyone considers valuing the other’s welfare to be a basic antecedent of empathic concern. Many focus on perceived similarity instead. For example, Nussbaum (2001) refers to Aristotle’s analysis of pity in the Rhetoric (1932). She suggests that for Aristotle the perceiver’s belief that his or her possibilities are similar to those of the person in need is a requirement for experiencing pity. Aristotle’s argument is that one’s own vulnerability to the same suffering is necessary in order to appreciate the reality of the other’s suffering, i.e., to perceive need. (For similar views among psychologists, see Davis, 1994, pp. 13–15; Gruen & Mendelsohn, 1986; and Houston, 1990). Nussbaum deviates from Aristotle, however, arguing that perceived similarity is important because to recognize one’s own related vulnerability makes the other’s suffering apparent and, potentially, relevant to one’s own value structure. Thus, she places the priority on valuing. Similarity is not necessary, but it may increase the likelihood one will recognize and care about the other’s suffering (see Nussbaum, 2001, especially Chapter 6).

My analysis parallels Nussbaum’s (2001) here; I also do not think similarity is necessary. If the other is thought to be sentient, then perceiving the other as in need and valuing the other’s welfare are sufficient conditions for the perceiver to appreciate the reality of the other’s suffering and to feel empathic concern. Even those of us who experience no fear or anxiety in close spaces can feel for someone we care about (value) who is claustrophobic. Awareness of this person’s discomfort and distress (perception of need) can be sufficient to evoke strong empathic feelings of sympathy and compassion.

For me to question the importance of perceived similarity as an antecedent to empathic concern may seem surprising. After all, I and others have used experimental manipulations of perceived similarity to induce empathic concern in research participants (p.43) (e.g., Batson, Duncan, Ackerman, Buckley, & Birch, 1981; Krebs, 1975; Stotland, 1969). However, a close look at the empirical evidence suggests that the effect of these similarity manipulations on empathic concern is not due to perceived similarity itself but to consequences of perceived similarity, such as liking, that reflect valuing of the other’s welfare (Batson, Lishner, Cook, & Sawyer, 2005; Batson, Turk, Shaw, & Klein, 1995; Lishner, 2003). For example, Batson, Turk et al. (1995, Experiments 1 & 2) found that an experimental manipulation of similarity led to increased valuing of a peer’s welfare. They also found that when the peer was in need, empathic concern was more clearly associated with this increased valuing than with perceived similarity.

In the most direct test to date of the effect of similarity on empathic concern, Batson, Lishner et al. (2005) found that female undergraduates reported as much empathic concern for a 40-year-old clothing-store clerk undergoing rehabilitation for a severely broken leg as they reported for a female undergraduate from their own university undergoing the same rehabilitation. This was true even though they thought the other undergraduate was much more similar to themselves than was the clerk. In a second and more extreme test of the effect of similarity, Batson, Lishner et al. (2005) found significantly more empathic concern for a child or a dog undergoing the same rehabilitation than for the undergraduate even though, once again, research participants thought the undergraduate was far more similar to themselves. (For other evidence that similarity may facilitate but is not a necessary condition for empathic concern, see Batson, Sympson, Hindman, Decruz, Todd, Weeks, Jennings, & Burris, 1996; Hodges, 2005; Hodges, Kiel, Kramer, Veach, & Villanueva, 2010; and Hygge, 1976).

Why Not Perspective Taking?

Even more surprising may be my focus on valuing the other’s welfare rather than on perspective taking. There is considerable evidence from laboratory research that perspective taking can increase empathic concern (e.g., Coke et al., 1978; Stotland, 1969; Toi & Batson, 1982), even concern for whales (Shelton & Rogers, 1981). Also in the natural stream of behavior, empathy may be increased by perspective taking instructions, including self-instructions (e.g., “Imagine what she’s going through!”). So, it may seem that perspective taking is the more basic construct, not valuing. For a long time, I thought it was. In discussions of the empathy-altruism hypothesis, I named (a) perceiving the other as in need and (b) adopting the other’s perspective as the two key antecedents of empathic concern (e.g., Batson, 1987, 1991; Batson & Shaw, 1991a). Clearly, the present formulation is different. Why the change?

There are three reasons. First, it is now apparent that a person can adopt another’s perspective—can actively imagine how the other thinks and feels about his or her situation (including a situation of need)—and still feel relatively little empathy. This can occur if one places either no value or negative value on the other’s welfare, and so has a dispassionate or hostile orientation toward the other. For example, Batson, Polycarpou, Harmon-Jones, Imhoff, Mitchener, Bednar, Klein, and Highberger (1997) found that even though those led to adopt the perspective of a convicted murderer serving a life sentence (p.44) reported more empathic concern for him than did those not led to adopt his perspective, they reported far less empathy than is typically reported in studies in which participants adopt the perspective of a stranger in need. Batson, Eklund, Chermok, Hoyt, and Ortiz (2007) found that less empathic concern for a young man hit by a car was reported by research participants led to place low value on his welfare than by participants led to place high value on his welfare. This effect of valuing was found even for those who adopted the young man’s perspective.

Second, it is now clear that a person can feel empathic concern for someone in need without being instructed to adopt the other’s perspective. Most people naturally place at least a moderate value on the welfare of other people—even total strangers—as long as there are no grounds for antipathy. Psychopaths are, of course, a conspicuous exception to this rule, but they comprise only a small percentage of the population. As a result of this moderate valuing, when research participants provided with no perspective-taking instructions learn about a stranger in clear, legitimate need, they typically report levels of empathic concern only slightly below the levels reported by those instructed to adopt the other’s perspective (Batson, Eklund et al., 2007). In such situations, it seems more accurate to say that adopting an objective perspective reduces empathic concern than to say that adopting the other’s perspective increases it. A moderately sympathetic orientation, not a dispassionate orientation, seems to be the default.

Third, Batson, Turk et al. (1995, Experiment 4) found that when participants learned (via false physiological feedback) that they felt empathic concern for a person in need, their valuing of this person’s welfare increased. Consistent with the link between values and emotions outlined at the end of Chapter 1, these participants appeared to make a backward inference from awareness of their empathic emotion to valuing (“If I feel empathic concern for her, I must value her welfare”), suggesting that they considered valuing necessary for empathy. Importantly, after the need was removed, empathic concern disappeared but the valuing remained, reflecting the more enduring character of valuing.

For these three reasons, I now think it best to focus on (a) perceiving the other as in need and (b) valuing the other’s welfare as the two key antecedents of empathic concern. In the flow of everyday life, perspective taking lies a little downstream from valuing the other’s welfare, on the diagonal line from valuing to the junction with perceiving need in Figure 2.1. Supporting this suggestion, Batson, Eklund et al. (2007, Experiment 2) found that increased valuing of another’s welfare led to the spontaneous adoption of an imagine-other perspective, which in turn led to increased empathic concern.

The downstream location of perspective taking explains why it can effectively induce empathic concern for someone in need. Even in the absence of prior valuing, it activates the valuing path. Indeed, use of perspective-taking instructions to induce empathy is usually a better research strategy than reliance on valuing. Valuing of another’s welfare is most likely to occur in close and enduring relationships (e.g., family relationships, friendships). Because such relationships are ongoing, with a past and a future, many motives not produced by empathic concern come into play. One can be motivated to benefit the other in order to reciprocate past benefits, to encourage reciprocation in the future, or in anticipation of being held accountable. The presence of these additional motives can (p.45) obscure any attempt to determine the nature of the motivation—altruistic or egoistic—produced by empathy.

Given these motivational confounds, it seems quite appropriate to use perspective-taking instructions rather than valuing to induce empathic concern in the laboratory. At the same time, it is important to recognize that in life outside the lab such instructions are not necessary to induce empathic concern. Valuing the other’s welfare produces a sympathetic orientation that naturally involves imagining how the other thinks and feels about events—perspective taking—and, coupled with perceived need, evokes empathic concern.

Intrinsic, Not Extrinsic Valuing

The type of valuing of another’s welfare that evokes empathic concern is what has been called intrinsic or terminal valuing, not extrinsic or instrumental valuing (Rokeach, 1973). The other is valued in his or her own right, not for what he or she may be able to provide.

When I perceive another whom I extrinsically value to be in need, I may feel concern, anxiety, fear, or sorrow, but these emotions are apt to be self-oriented—evoked by the implications of the other’s plight for my own welfare. I may be upset if I hear that the mechanic who promised to have my car ready on Tuesday has come down with the flu, and my car will not be ready until Friday. If I am honest, however, I may also admit that I am upset almost entirely (if not entirely) about the delay in getting my car. I give little thought to—and have little feeling for—the discomfort and difficulties the mechanic is experiencing.

In contrast, learning that someone whom I intrinsically value is sick is likely to cause me to feel sympathy, compassion, anxiety, or sorrow for this person. Because I have incorporated the other’s welfare into my own value structure, I imagine how he or she is affected by the situation—I spontaneously perspective take—and feel empathic concern. It is the threat to his or her welfare, not to my own, that evokes my emotional response.

All this assumes, of course, that it is possible to value another’s welfare intrinsically. There seems little doubt that we can value another’s welfare extrinsically, even someone quite close. A young child may be upset that his mother is sick because of the implications for his own welfare (Knafo, Zahn-Waxler, Van Hulle, Robinson, & Rhee, 2008; Zahn-Waxler, Radke-Yarrow et al., 1992). So may an adult who is faced with the illness or injury of a spouse. Extrinsic valuing of the other underlies interdependence theory analyses of close relationships, which assume that each person values the relationship to the extent that the partner is necessary for the person’s own well-being (Berscheid, 1983; Kelley, 1979).

Can we ever value another’s welfare intrinsically? There are reasons to think that we can. First, it is important to note that extrinsic valuing and intrinsic valuing are not mutually exclusive. What was once valued extrinsically may, with time, become functionally autonomous and valued in its own right (Allport, 1937). This is true even though perceptions of extrinsic valuing can undermine perceptions of intrinsic valuing. (Reflecting such undermining, I would probably have felt more empathic concern on hearing my (p.46) mechanic had the flu were my car not in the shop at the time—see Aronson & Carlsmith, 1963; Lepper, 1983.) Other terms for what I am calling intrinsic valuing of the other’s welfare are caring, loving, or being close.

When one person values, cares for, loves another—for example, when a mother loves her child—there are likely to be feelings of heartache and sadness at prolonged separation, and feelings of warmth and joy at reuniting. Cognitive processes such as perceived similarity, familiarity, and attractiveness can contribute to love. However, its basic character seems to be affective and evaluative. Like the related but more general concepts of attitude and sentiment, love involves a relatively enduring value placed on the target, even though love can, of course, end. Love is often thought to be an emotion, but it may be more appropriate to think of love as a form of valuing. Threats to the welfare of a loved one can evoke a range of emotions, including empathic concern.

Linking Valuing the Other’s Welfare and Empathic Concern to Parental Nurturance

The parent’s valuing of and care for the child almost certainly has a genetic base (see Bell, 2001; Bowlby, 1969; Hoffman, 1981a). Less certain but certainly intriguing, the genetically based caring of parent for child may provide a biological substrate for all intrinsic valuing of another’s welfare and, thereby, for all empathy-induced altruism in humans.

Today, parental nurturance is rarely mentioned as the evolutionary origin of empathy-induced altruism, but it was frequently mentioned a century ago. At that time, psychologists were strongly influenced by Darwin (1871), who spoke of instinctive love based on parental and filial affections and linked it to “the all important emotion of sympathy” (p. 308). The suggestion that parental nurturance may be a source of empathic concern and of altruistic motivation has recently resurfaced outside psychology in the writings of primatologist Frans de Waal (1996) and of philosopher Elliott Sober and biologist D. S. Wilson (1998).

Is it possible that feelings of tenderness and compassion, even for strangers, are grounded in the strong impulse for mammalian parents to provide care for their vulnerable and dependent offspring? It certainly seems clear that if mammalian parents were not intensely interested in the welfare of their young—so interested as to put up with endless hassles, exhaustion, and even risks to their personal safety—these species would quickly die out (also see Bartels & Zeki, 2004; Bell, 2001; Hoffman, 1981a; MacLean, 1990; Taylor, 2002; Zahn-Waxler & Radke-Yarrow, 1990).

McDougall’s Parental Instinct and Tender Emotion

William McDougall (1908) provided what is perhaps the most systematic argument to date for parental nurturance as the basis for empathy-induced altruism, even toward (p.47) strangers. He described the “parental instinct,” which he considered to be the most powerful of all instincts, and associated “tender emotion.” McDougall did not think of instincts as automatic, reflexive responses. For him, all instincts included a cognitive, an affective, and a motivational component. The cognitive and motivational components were modifiable by experience and learning, but the affective component was not; it defined the character of the instinct. The tender emotion defined the character of the parental instinct. According to McDougall (1908), this instinct

is primarily to afford physical protection to the child, especially by throwing the arms about it; and that fundamental impulse persists in spite of the immense extension of the range of application of the impulse…. Tender emotion and the protective impulse are, no doubt, evoked more readily and intensely by one’s own offspring, because about them a strongly organized and complex sentiment grows up. But the distress of any child will evoke this response in a very intense degree in those in whom the instinct is strong…. By a further extension of the same kind the emotion may be evoked by the sight of any very young animal, especially if in distress…. In a similar direct fashion the distress of any adult (towards whom we harbor no hostile sentiment) evokes the emotion. (pp. 61–63)

McDougall’s tender emotion is clearly empathic as I am using that term. And, consistent with the empathy-altruism hypothesis, McDougall believed that the motivation evoked by the tender emotion was altruistic:

From this emotion and its impulse to cherish and protect spring generosity, gratitude, love, pity, true benevolence, and altruistic conduct of every kind; in it they have their main and absolutely essential root without which they would not be. (McDougall, 1908, p. 61)

Thus, for McDougall, tender, empathic feelings and altruistic motivation are key components of the parental instinct in humans, which can be generalized not only to other children but also to adults in need.

We should not accept such an idea too quickly. Many mammalian species lack the prefrontal cortical structures and cognitive abilities necessary to experience tender, empathic feelings. Yet these species display parental care. For McDougall, human parental nurturance involves (a) inference about the internal states of others, (b) perception of need, (c) intrinsic valuing, (d) empathic concern, and (e) altruistic motivation. If McDougall is right, the human parental instinct goes well beyond nursing, providing other kinds of food, protecting, and keeping the young close—the activities that characterize parental care in most other mammalian species. It includes feeling for the child based on inferences about the desires and feelings of the child (“Is that a hungry cry or a wet cry?” “She won’t like the fireworks; they’ll be too loud.”). It also involves a clear recognition of the distinctiveness, even possible dissimilarity, of self and other. Parents must recognize that a child’s needs may be quite different from their needs. They must also recognize that the child’s capacity to deal with needs may be quite different from their own.

Humans have doubtless inherited key aspects of their parental instinct from ancestors they share with other mammalian species, but in humans this instinct has become considerably less automatic and more flexible. Parental care based on the tender-emotion (empathic concern) did not replace the more primitive hard-wired stimulus-response circuits. (p.48) It supplemented them, increasing the flexibility with which they are employed (Bell, 2001; A. Damasio, 2002; MacLean, 1990; Sober, 1991; Sober & Wilson, 1998; Taylor, 2002; Zahn-Waxler & Radke-Yarrow, 1990).

Note that McDougall’s grounding of the parental instinct in the tender emotion is quite different from attachment theorists’ ethological grounding of what they call “the caregiving behavioral system.” Their ethological perspective leads attachment theorists to focus on caregiving as cue-based and reactive rather than as proactive, cognitively flexible, and emotionally mediated (see Bowlby, 1969; George & Solomon, 1999; but also see Bell, 2001). McDougall’s focus on other-oriented tender emotion is also quite different from de Waal’s (2009) grounding of the parental instinct in emotional contagion and affective resonance.

Further, for McDougall, the human parental instinct and associated tender emotion have a range of applicability that extends well beyond parent-child relations. Through cognitive generalization based on learning and experience, this instinct and emotion come into play in many if not all cases of intrinsic valuing of another’s welfare, as well as in the resulting empathic concern and altruistic motivation. Both women and men must be capable of caring about the welfare of non-kin—even strangers—in something like the same way, if not the same degree, that they care for their own children. Moreover, the parental instinct and associated tender emotion must not be limited to those who have children; it must be operative from an early age. Is such a view at all plausible? Recent research suggests that it is plausible, albeit far from certain.

Neurophysiology of Parental Care and Empathic Concern

First, there is research that looks at the neurophysiology of brain regions related to parental care, empathic concern, and altruism. Although to date only limited work has been done, several luminaries of neurophysiology have provided general perspectives.

Paul MacLean (1990) described the human brain as triune, consisting of a hierarchy of “three-brains-in-one”: Oldest is an evolutionarily ancient reptillian (or protoreptillian) brain that we share with reptiles and mammals. Next is a paleomammalian brain (containing the limbic system) that we share with all mammals. Most recent is a neomammalian brain (frontal and prefrontal cortex) that is found only in higher mammals and that reaches its greatest proportions in humans. According to MacLean (1990), each of these brains has “its own special intelligence, its own subjectivity, its own sense of time and space, and its own memory, motor, and other functions” (p. 9). Each operates somewhat independently but is not completely autonomous. The three intermesh and function together. For MacLean, parental care, play, and social bonding—“functions that would seem to have favored the evolution of the human sense of empathy and altruism” (1990, p. 520)—all arise from the interrelationship of areas of the neomammalian frontal cortex with a subdivision of the paleomammalian limbic system.

At a more detailed level, Antonio and Hanna Damasio have sketched the implications for the neurophysiology of empathy of their research with patients who have brain lesions. (p.49) In addition to the limbic areas involved in the expression of emotion (MacLean, 1990), Antonio Damasio proposed that “the brain regions in the metarepresentation of mental states are critical for the process of empathy, and they include… regions of parietal association cortex and of prefrontal cortex” (2002, p. 269; also see Decety & Chaminade, 2003; Eslinger, 1998; Immordino-Yang, McColl, Damasio, & Damasio, 2009; Kim, Kim, Kim, Jeong, Park, Son, Song, & Ki, 2009; Lamm, Batson, & Decety, 2007; and Ruby & Decety, 2004). Similarly, Hanna Damasio concluded:

There is a system in certain sectors of the prefrontal cortex that is critical for the learning and maintenance of certain aspects of social behavior that pertain to interpersonal relationships. After damage to this system, empathy, as well as emotions such as embarrassment, guilt, pride, and altruism, are not evoked, and personal and social decisions become defective….

It is possible, and indeed likely, that some of the adaptive interpersonal behaviors that were lost in these patients are actually preset in neural systems that include prefrontal components…. Without certain sectors of the prefrontal cortex, empathy, along with other adaptive social behaviors, becomes impaired. (2002, pp. 281–282)

Behavioral observations seem quite consistent with these suggestions about the neurophysiology of empathic concern, as well as with the neurophysiological evidence that for normal operation of the human parental instinct, cortical processes attentive to the internal state and needs of the child interact with midbrain-based reaction tendencies to produce goal-directed motivation (Damasio, 1994; MacLean, 1990).

Antonio Damasio (1994, 1999, 2003) has repeatedly pointed out that one of the virtues of relying on emotions and goal-directed motives to guide action—rather than relying on more automatic stimulus-response patterns (his “regulatory mechanisms”)—is that emotions and their associated goal-directed motives can be adaptive under a wide range of environmental conditions, circumstances, and events. Such flexibility seems highly desirable when caring for human offspring. To illustrate the flexibility that emotions introduce with an emotion quite different from empathic concern, consider anger. Aggressive responses occur in many species that likely do not experience anything like the emotion we call anger. Among humans, however, aggressive responses are stimulated, tempered, and generalized by feelings of anger that are a product of complex cognitive assessment of the situation, including assessment of the intentions of others. Similarly, tender, empathic feelings permit more flexible and adaptive parental care, care that is not simply reflexive or reactive to distress cues but is directed toward the goal of enhancing the child’s welfare as needed in the particular situation. This flexibility includes anticipation and prevention of needs, even evolutionarily quite novel ones—such as the need to avoid sticking a pin in an electrical outlet.

Bell (2001) reached a similar conclusion, describing the human parental caregiving system as “directed toward the needs of the infant” (p. 220) and therefore based on relatively sophisticated cognitive processes. At the same time, he recognized that this system is linked to “some emotional processes located in older parts of the brain that appear to follow a different, emotional logic” (Bell, 2001, p. 216). Bell assumed that this sophisticated and flexible caregiving system is present not only in humans but also in other primates. As already noted, Tomasello (1999; Call & Tomasello, 2008) and Povinelli (p.50) (Povinelli et al. (2000; Penn et al., 2008) give us reason to reserve judgment on such an assumption.

Consistent with (but certainly not conclusively supporting) this depiction of human parental care, Bartels and Zeki (2004) found that mothers looking at photos of their own child compared to photos of another child (aged-matched) showed increased activation of the periaqueductal gray (PAG) region in the midbrain, which is known to be involved in maternal behavior in mammals, including humans. They also found increased activation of regions in the cortex associated with higher cognition and emotion (especially positive, tender feelings) as well as with goal-directed activity—the medial insula, dorsal and ventral anterior cingulate cortex, lateral orbito-frontal cortex, and lateral prefrontal cortex. The first three of these regions, all associated with emotion and motivation, are known to have direct connections with the PAG. (For informative analyses of the relevant neuroanatomy, see Allman et al., 2005, and Craig, 2005.)

Neurochemistry of Parental Care

Shelley Taylor (2002) proposed the existence of a “tending instinct” that is remarkably similar to McDougall’s parental instinct, even though she made no reference to his work. Taylor suggested that the instinct to tend and nurture offspring and to establish a range of attachment bonds (i.e., an instinct to “tend and befriend”) underlies “various forms of tending throughout society” (Taylor, 2002, p. 158). She also suggested that this instinct may have its neurochemical base in the neuropeptide hormone oxytocin and the endogenous opioid peptides (EOPs). There is some evidence, which Taylor admitted is limited and spotty, that oxytocin may be released not only during sexual intercourse, at birth, and during nursing, but also in other affiliative experiences (see, for example, the evidence provided by Turner, Altemus, Enos, Cooper, & McGuinness, 1999; reviews of some of this evidence are provided by Bell, 2001, and Panksepp, 1998, as well as by Taylor, 2002). More recently, Feldman, Weller, Zagoory-Sharon, and Levine (2007) reported an association between mother’s level of plasma oxytocin and cognitive as well as behavioral aspects of mother-infant bonding, including thoughts about the infant and vigilance for the infant’s welfare.

Research on oxytocin is certainly intriguing. It might lead one to think that oxytocin provides a neurochemical link between parental care and other forms of care, including empathy-induced altruism. But this research is also complex, and results are neither consistent nor yet clear. Oxytocin has been found to be highly associated with maternal care and social attachment in some mammalian species but not in others—sometimes even closely related ones. There is, for example, an association in rats but not in mice (Carter, 1998; Insel, 2000; Kendrick, 2000; Nelson & Panksepp, 1998; Olazábal & Young, 2006). And vasopressin, not oxytocin, may underlie paternal care and pair-bonding in at least some mammalian species (Curtis & Wang, 2003; Insel, 1997, 2002). Those most knowledgeable about research on oxytocin suggest that it is still too soon to make any strong claims about the neurochemistry of parental care and attachment in humans (p.51) (e.g., Carter, 1998; Donaldson & Young, 2008; Insel, 2000, 2002; Panksepp, 1998). As Insel (2000) summarized,

The available data support the hypothesis that oxytocin is critical for maternal behavior and pair-bond formation in select nonhuman animals. Humans have oxytocin and brain oxytocin receptors, but the role of this neuropeptide system in human attachment remains highly speculative. (p. 176)

Evidence from human oxytocin studies over the past few years may have tipped the scales from highly speculative to highly suggestive, but it is still too soon to reach a verdict.

Evidence Regarding Generalization of Tender Feelings and Nurturant Care

What about generalization beyond progeny? McDougall (1908) claimed that we can extend the tender feelings and nurturant care that emerged as part of the human parental instinct to a wide range of others, including adult strangers and even members of other species (especially pets). The suggestion is that through cognitive generalization we “adopt” non-progeny, making it possible for their needs to evoke empathic concern and altruistic motivation (Batson, 1987; Hoffman, 1981a). The prospect of such generalization may seem implausible and at odds with the theory of natural selection, as was argued by Boehm (1999).

However, it is important to recognize that genetically hardwired parental care need not be progeny specific to be effective. Insel (2002) has noted that “rat mothers will show intense devotion and defense of their young, but they are not selective in their maternal behavior, offering the same level of care to unrelated young in the nest” (p. 255). Presumably, occurrence of unrelated young in a rat’s nest is sufficiently rare that there has not been strong selection pressure for a more discriminating maternal response. Nor is adoption rare in other mammalian species. What is currently discussed as alloparenting and cooperative breeding is found in a range of primate species, including humans, as well as in elephants, canids (wolves, dogs), rodents, a number of bird species, and, of course, the social insects (Hrdy, 2009).

In a highly interdependent and cooperative species like our own, natural selection may not only have tolerated generalized parental nurturance; there may actually have been a selective advantage to extending the genetically hardwired nurturant impulse beyond one’s own offspring. Due to selection pressure on the small, closely knit hunter-gatherer bands in which our genetic predispositions for social behavior are thought to have evolved (Caporeal, Dawes, Orbell, & van de Kragt, 1989; Hrdy, 1999, 2009; Kelly, 1995), generalization of the impulse to provide nurturant care for our own offspring to include care for younger siblings (Dunn & Kendrick, 1982; Hrdy 2009), care for the offspring of other band members, and even care for other adults in the band may have increased the likelihood of our genes surviving (Sober & Wilson, 1998). And, to the (p.52) extent that the human nurturant impulse relies not on cue-based stimulus-response patterns but on cognitively based other-oriented emotions such as tender, empathic concern, it would be relatively easy to generalize.

Within contemporary society, the prospect of such generalization appears more plausible when one thinks of the tender care typically provided by nannies and workers in day-care centers, by adoptive parents, and by pet owners. One can even see tender, nurturant care provided by young children to people, pets, stuffed animals, and dolls—and by pets to family members (Hoffman, 1981a; Zahn-Waxler & Radke-Yarrow, 1990). Clearly, tender, nurturant feelings are felt not only by mothers, by parents, by women, or by adults. As early as the second year of life, children of both genders have them. The parental instinct is deep and pervasive—although receipt of nurturance seems necessary for its normal expression (Harlow, Harlow, Dodsworth, & Arling, 1966; Hrdy, 2009).

Consistent with the prospect of generalization, some have proposed that parental nurturance may play a role in adult friendships and love relationships. Curtis and Wang (2003) reflected as follows on research over the previous decade concerning the role of oxytocin in pair bonding in monogamous prairie voles (in contrast to non-monogamous meadow voles):

One possibility for the origin of pair bonding is that pair-bonding species have co-opted the mechanism (or mechanisms) by which maternal bonds are formed. This possibility is further supported by observations that even sexually naïve male prairie voles display maternal-type behaviors when exposed to pups, and that prairie vole mothers display considerably more maternal care than do meadow vole mothers. (p. 51)

Grewen, Girdler, Amico, and Light (2005) reported an association between closeness of romantic relationship (reflected in partner support) and level of plasma oxytocin before and after warm contact with the partner. This association was found for both men and women.

Evidence of plasma oxytocin increase has also been reported in affiliative experiences of humans (men and women) with dogs (Odendaal & Meintjes, 2003). And oxytocin infusion (nasal) was reported to increase generosity toward strangers among men in a competitive situation—an Ultimatum Game (Zak, Stanton, & Ahmadi, 2007). Zak et al. (2007) speculated that the increase in generosity was mediated by the effect of oxytocin on empathic emotion. Finally, Barraza and Zak (2009) found that after watching a video interview in which a father describes the plight of his 2-year-old son who has a terminal brain tumor, including scenes of the child in the hospital, UCLA undergraduates not only reported more empathic concern but also showed increased plasma oxytocin.

Turning from neurochemistry to neurophysiology, Singer et al. (2004) found increased activation in the anterior insula and rostral anterior cingulate cortex among women informed that their romantic partner was receiving a painful (vs. not painful) electrical stimulation. Bartels and Zeki (2000) found increased activation of many of the same cortical regions (but different midbrain regions) when participants looked at a photo of their romantic partner compared to looking at photos of friends. Jackson et al. (2006) found similar activation among participants imagining an unfamiliar other in a painful (vs. not (p.53) painful) situation. Finally, Lamm et al. (2007) found that activation of the medial anterior cingulate cortex correlated positively with self-reported empathic concern among participants watching unknown patients undergo a painful therapeutic treatment. In sum, several of the same (or closely associated) regions activated in mothers looking at photos of their own child (Bartels & Zeki, 2004) are activated when cued to a loved one’s distress or when seeing—or even imagining—an unknown other in distress (also see Immordino-Yang et al., 2009; Kim et al., 2009).

Also consistent with the idea of generalized parental nurturance, Batson, Lishner et al. (2005) found that a child, dog, or puppy in need evoked more empathic concern than did a fellow university student with exactly the same need. And Lishner, Oceja, Stocks, and Zaspel (2008) found that empathic concern for adults in need is enhanced when the adult has a more infant-like face or voice.

Thus, there is a range of evidence consistent with the idea that parental nurturance may provide a biological substrate for intrinsic valuing of another’s welfare and for empathy-induced altruism in humans. Although not conclusive, the existing evidence supports the plausibility of the idea that four evolutionary developments may underlie the human capacity to care for the welfare of both progeny and non-progeny as an end in itself, not simply as an instrumental means of caring for one’s own welfare. The first development is the evolution in mammals of parental nurturance (Bell, McDougall). Second is the evolution in humans and possibly a few other species of the ability to see others as sentient, intentional agents and, thereby, to recognize other’s needs, even subtle ones (Povinelli et al., Tomasello). Third is the evolution of tender, empathic emotions as an important component of parental nurturance (Bell, Darwin, McDougall). Fourth is the evolution of cognitive capacities that make it possible to generalize valuing of another’s welfare and tender, empathic feelings beyond offspring (McDougall).

If this analysis is correct, then we have an answer to the question raised earlier about whether the capacity to value another’s welfare intrinsically is a violation of the principles of natural selection. It is not. Parental nurturance is entirely consistent with those principles. To value another’s welfare intrinsically is, however, in clear violation of a central tenet of standard versions of the theory of rational choice—exclusive interest in one’s own welfare. If humans are able to value another’s welfare intrinsically, then versions of rational choice that assume all human behavior is directed toward maximizing one’s own welfare need radical revision (see Chapter 9; also Batson & Ahmad, 2009a).

A New (Actually Old) Evolutionary Perspective on Altruism in Humans

It is important to note that generalized parental nurturance is different from the evolutionary biologists’ idea of inclusive fitness (kin selection) as a genetic basis of altruism (Hamilton, 1964). The former refers to a specialized and specific adaptation—an instinct—whereas the latter proposes a general principle. A genetic impulse to care for (p.54) offspring certainly falls within the purview of inclusive fitness. On average, half the offspring’s variable genes (the less than 1 percent of our genes that vary among humans) are one’s own, so to care for offspring increases the likelihood of one’s variable genes surviving, enhancing inclusive fitness. Offspring are not, however, an indirect way to get one’s genes into the next generation, which is the issue addressed by the idea of inclusive fitness. Offspring are one’s genes in the next generation. As a result, parental care does not speak to the problem that Hamilton (1964) was trying to solve with the concept of inclusive fitness, the problem of apparent evolutionary altruism.

Recall Sober and Wilson’s (1998) distinction between evolutionary altruism and psychological altruism presented in Chapter 1. Care for offspring is evolutionary egoism, not evolutionary altruism. That is, it is a case of an organism acting in a way that increases its own personal reproductive fitness. For this reason, when citing examples of inclusive fitness, evolutionary biologists focus on care for siblings or more remote kin; they almost never mention parental care of offspring. (Parental care is more likely to get discussed under the topic of “parental investment”—the jockeying between parents in which each tries to ensure that their joint offspring survive through the least possible expenditure of his or her own time and energy; see Buss & Kenrick, 1998; Trivers, 1972.)

Especially in higher mammals, the effects of parental nurturance on reproductive fitness are much more focused, direct, and straightforward than are the effects of inclusive fitness. Because offspring of higher mammals are unable to fend for themselves for an extended period after birth, there is strong selection pressure to develop a mechanism that leads parents to provide care. It is far less clear that an impulse to care for siblings and more remote kin—behavior often attributed to inclusive fitness—would receive strong selection pressure.

One can build a clear case for a genetically hardwired impulse to care for siblings in the social insects, where sisters share three-fourths of their genes and are themselves sterile (Hamilton, 1964; yet see Wilson, 2005, and Wilson & Wilson, 2007, for doubts). One can also build a clear case for such an impulse in the naked mole rat, a mammalian species with a sterile worker caste (Sherman, Jarvis, & Alexander, 1991). But in humans, the case for a degree-of-kinship-based genetic impulse of the sort Hamilton (1964) postulated is far from clear (see Campbell, 1975, and Batson, 2010). Data most frequently cited as support (e.g., Burnstein, Crandall, & Kitayama, 1994; Essock-Vitale & McGuire, 1980) are easily amenable to alternative explanations and are at least as compatible, if not more so, with generalized parental nurturance as with inclusive fitness (see especially Korchmaros & Kenny, 2001).

In a species like ours, in which each normally developing individual has the potential to procreate and, thereby, to place his or her genes directly in the next generation, genetic selection for use of an indirect route through helping others proportional to degree of kinship is not likely. Helping kin is far more likely to be a product of social norms and cultural mores—cultural evolution (Campbell, 1975; Richerson & Boyd, 2005). In contrast, it is hard to doubt the existence of a strong, genetically hardwired impulse for parental care in humans—McDougall’s parental instinct—and there is good reason to believe that empathic concern—McDougall’s tender emotion—plays an important role in the (p.55) expression of this impulse. This parental instinct is strong but flexible. It can be overridden in certain circumstances (which, when extreme, can produce abandonment and even infanticide—Hrdy, 1999; Soltis, 2004; Zeifman, 2001). There is also good reason to believe that it can generalize beyond progeny.

If we wish to speculate about genetic bases for human altruism, as many clearly do, then I think we are on far firmer ground—both logical and empirical—if we focus on cognitive generalization of tender, empathic feelings that emerged to add flexibility to a genetically hardwired parental instinct (McDougall, 1908) than if we focus on a genetically hardwired impulse toward inclusive fitness (Hamilton, 1964), reciprocal altruism (Trivers, 1971, 1985), or some combination of the two (e.g., Brown & Brown, 2006)—or even on genetically hardwired impulses toward sociality, cooperation, trust, and coalition formation (Caporeal et al., 1989; de Waal, 1996; Frank, 2003; Sober & Wilson, 1998).

Ideas of inclusive fitness, reciprocal altruism, and group selection have dominated recent thought about natural selection and the genetic basis of altruism. The possibility that parental nurturance might serve as a genetic substrate for human altruism has been largely ignored. With Sober and Wilson’s (1998) distinction between evolutionary altruism and psychological altruism, perhaps it will once again be possible to recognize the importance for psychological altruism of a process like parental nurturance, which does not qualify as evolutionary altruism. Certainly, it seems far from coincidental that Sober and Wilson (1998) ended their book with a consideration of parental nurturance as a plausible example of psychological altruism.

Individual Differences as Moderators of Empathic Concern

My analysis of the antecedents of empathic concern has focused on two factors—perceiving the other as in need and valuing the other’s welfare. I have not considered individual differences. I have not, even though I think the strength of empathic feelings is affected by individual difference factors—some of which may be genetically hardwired (Emde, Plomin, Robinson, Corley, DeFries, Fulker, Reznick, Campos, Kagan, & Zahn-Waxler, 1992; Knafo et al., 2008). The reason I have not considered these individual difference factors is that I believe they are more appropriately thought of as moderators of the effect of perception of need and valuing on empathic concern, not as direct antecedents.

Dispositional Empathy?

But, one may ask, is there not a specific disposition to experience empathy? A number of self-report questionnaire measures of such a disposition have been developed—most prominently, Davis’s (1983) Interpersonal Reactivity Index and Mehrabian and Epstein’s (1972) Questionnaire Measure of Emotional Empathy. However, I believe that there is reason to doubt the validity of these measures, which ask people to report whether, for example, “I often have tender, concerned feelings for people less fortunate than me” (p.56) (an item from the Empathic Concern scale of Davis’s Interpersonal Reactivity Index). One might strongly agree with such a statement because it is true, but one might also agree because one erroneously believes it is true, because one wants to believe it is true, or because one wants others to believe it is true. Each of these reasons for agreement is, under certain circumstances, likely to be associated with increased helping of those in need. But only the first reason provides a valid indication of a disposition to experience empathic concern. So, according to the empathy-altruism hypothesis, only the first reason should be associated with altruistic motivation to help.

There is considerable evidence that scores on Davis’s Empathic Concern scale correlate positively both with reports of empathic concern and with helping in a range of situations (see Davis, 1994, for a review)—and perhaps even with increased activity in brain regions associated with the experience of affect (e.g., Singer et al., 2004; but also see Decety, 2010b; Vul, Harris, Winkielman, & Pashler, 2009). However, consistent with the concern about validity, there is also evidence that scores on the Empathic Concern scale and other measures of dispositional empathy are associated, not with altruistic motivation but with an egoistic desire to see oneself as altruistic (Archer, Diaz-Loving, Gollwitzer, Davis, & Foushee, 1981; Batson, Bolen, Cross, & Neuringer-Benefiel, 1986).

Eisenberg and her colleagues (Carlo, Eisenberg, Troyer, Switzer, & Speer, 1991; Eisenberg, Miller, Schaller, Fabes, Fultz, Shell, & Shea, 1989) have challenged the claim that scores on the Empathic Concern scale are associated with egoistic rather than altruistic motivation to help, but the basis of their challenge is questionable. Their challenge relies on the results of two studies. In each, participants had a low-cost opportunity to help in response to a rather stereotypic need (in one study, a single mother with two hurt children; in the other, a recently assaulted young woman experiencing flashbacks). Given the nature of these help opportunities, it is difficult to know whether helping was motivated by an altruistic desire to relieve the need or by an egoistic desire to avoid social and self-censure for failure to help when norms dictate one should. For each study, Eisenberg and her colleagues made no claim to be able to make this essential distinction (see Carlo et al., 1991, p. 450; Eisenberg, Miller et al., 1989, p. 62).

To express doubt about the validity of self-report questionnaire measures of dispositional empathy is not to express doubt that individual differences affect the experience of empathy. Such effects clearly exist. In the same situation, some people feel more empathy than others. My doubt is about whether existing self-report questionnaires validly reflect these differences. There is too much room for social desirability and positive self-presentation in responses to these questionnaires.

More fruitful than using self-reports to measure individual differences in dispositional empathy are efforts to assess likely dispositional moderators of empathic responding in specific situations. Such moderators include general emotionality and the regulation of emotion (Davis, Luce, & Kraus, 1994; Eisenberg, Fabes, Murphy, Karbon, Maszk, Smith, O’Boyle, & Suh, 1994; Eisenberg, Losoya, & Spinrad, 2003; Wiesenfeld et al., 1984), as well as attachment style (level of anxiety about and desire to avoid social relations—Mikulincer et al., 2001; Mikulincer & Shaver, 2003; Mikulincer, Shaver, Gillath, & Nitzberg, 2005). Doubtless there are others as well.

(p.57) Gender Differences?

It has also been suggested that gender may be an important antecedent of empathic concern. Specifically, it has been suggested that women feel more empathy than men (de Waal, 2009; Hoffman, 1977). Evidence for this claim is not, however, especially strong, and it is largely limited to self-report measures of empathy (Eisenberg & Lennon, 1983) or to gender-appropriate expressions of concern (Zahn-Waxler, Robinson, & Emde, 1992). Such responses likely reflect gender differences in expectations about, and in the perceived desirability of, feeling and expressing empathic emotion.

Research to be reviewed in Part II provides much evidence that men are quite capable of experiencing empathic concern. Still, there may be genuine gender differences that moderate the experience of empathy. There is some evidence that women are, in general, more emotional than men, or at least are more emotionally expressive (Buck, 1984; Eisenberg et al., 1994; Zahn-Waxler, Cole, Welsh, & Fox, 1995). Zahn-Waxler, Robinson, and Emde (1992) even link possible gender differences in empathic concern to gender differences in parental nurturance in a manner reminiscent of McDougall (1908): “A biologically based explanation for the origins of gender differences in empathy would be consistent with the childbearing and child-rearing roles of females. Empathic caregiving is required if the infant is to survive and thrive” (Zahn-Waxler et al., 1992, p. 1045). Of course, moderation of empathic concern by gender could reflect socialization and culture as well.

Of critical importance for the present theory, there is as yet no evidence from any of the research conducted to test the empathy-altruism hypothesis (summarized in Chapters 5 & 6) that the empathy-altruism relationship differs by gender. Even if women are more likely to experience empathic concern, once empathic feelings are evoked they have similar motivational consequences for men as for women—at least when generalized beyond offspring.

A Return to the Seven Other Empathy-Related States

With this analysis of the proposed two antecedents of empathic concern before us, it may be useful to return to the seven other empathy-related states discussed in Chapter 1, examining how each relates to the two antecedents. Concepts 2, 3, and 4—adopting the other’s posture (motor mimicry), feeling as the other feels, and projecting oneself into the other’s situation—are different strategies for knowing what the other is thinking and feeling (Concept 1). Thus, each should affect perception of the other as in need. Concepts 3 and 4, under conditions specified earlier, may facilitate adoption of the other’s perspective (Concept 5), a product of and a research proxy for valuing the other’s welfare. None of the first four states affect valuing the other’s welfare directly.

I have suggested that in the natural flow of behavior, adopting another’s perspective (Concept 5) is a consequence of valuing the other’s welfare, not an antecedent. (p.58) However, in the absence of prior valuing, imagining another’s thoughts and feelings (i.e., an imagine-other perspective) can be induced directly through instructions. In this case, it—coupled with perception of need—should evoke empathic concern.

Concept 6, imagining how one would think and feel in the other’s place (i.e., an imagine-self perspective), may affect perception of the other as in need. This is especially likely if (a) there are no clear, independent cues about the other’s need, and (b) one has reason to believe that one’s own response to the situation and the other’s response would be similar—or would differ in a predictable way. In addition, adopting an imagine-self perspective may serve as a stepping-stone to adopting an imagine-other perspective (Concept 5), especially if the other’s situation is not also threatening to the self.

Concept 7, feeling vicarious personal distress, does not affect either antecedent of empathic concern. Rather, it is a self-oriented emotion evoked by perceiving the other as in need, an emotion that can be experienced alongside empathy. Feelings of personal distress are likely to produce egoistic motivation to reduce one’s own distress.

In sum, empathic concern is proposed to be a natural consequence of two factors, perceiving the other as in need and intrinsic valuing of the other’s welfare. Therefore, only these two antecedents are represented in Figure 2.1. Most of the other empathy-related states discussed in Chapter 1 can contribute to empathic concern indirectly through their effect on perception of the other as in need (Concepts 1, 2, 3, 4, and 6). None is an antecedent of valuing the other’s welfare, although Concept 5 refers to a key consequence of valuing that has also been used in research as a proxy for valuing—adopting the other’s perspective. Two of the empathy-related states (Concepts 4 and 6) may lead to adoption of the other’s perspective and, thereby, increase empathic concern. Parental nurturance may be the prototype and provide a genetic substrate for the human capacity to extend intrinsic valuing beyond oneself to others, even strangers.