REPORTS Altruistic Helping in Human Infants and Young Chimpanzees of b Felix Wareken"and Michael Tomasello 中 This requires asks It is not worthy that they did so in alm 丁 (849% helping acts tionally Coonitively to help someone achievinga wrong (coreetable) nding control task i which the same that this problem for the an primates are There an in cac on their prop sity to help used primates uman I age.I was ot tryn problem nfa receive any ard or r of studies have dem rated tha when it involves obie ts other than food be experiment and food 00 in the ex nieghtpnechdcethetr c( for exa comfortin dition for a betw the sam hasi are no stigated instrumental helpi study we ented 24 18- of problems used in child study Category Task Problem a goal.This varicty of Ou-of-r ach Marker The adult a essfully reaches goal and that he bu into it (experimentaD ersus bu his problems in reaching the goal Wrong result Book to on top of the stack Wrong means Flap temnatively heththetheonpupo small hole www.sciencemag.org SCIENCE VOL 311 3 MARCH 200 1301Altruistic Helping in Human Infants and Young Chimpanzees Felix Warneken* and Michael Tomasello Human beings routinely help others to achieve their goals, even when the helper receives no immediate benefit and the person helped is a stranger. Such altruistic behaviors (toward non-kin) are extremely rare evolutionarily, with some theorists even proposing that they are uniquely human. Here we show that human children as young as 18 months of age (prelinguistic or just-linguistic) quite readily help others to achieve their goals in a variety of different situations. This requires both an understanding of others’ goals and an altruistic motivation to help. In addition, we demonstrate similar though less robust skills and motivations in three young chimpanzees. Helping is an extremely interesting phe￾nomenon both cognitively and motiva￾tionally. Cognitively, to help someone solve a problem, one must know something about the goal the other is attempting to achieve as well as the current obstacles to that goal. Motivationally, exerting effort to help another person—with no immediate benefit to oneself—is costly, and such altruism (toward non-kin) is extremely rare evolutionarily. In￾deed, some researchers have claimed that humans are altruistic in ways that even our closest primate relatives are not. A powerful method to test this idea is to directly compare human infants and our closest primate relatives (chimpanzees) on their propensity to help. Such a comparison may enable us to distin￾guish aspects of altruism that were already present in the common ancestor of chimpan￾zees and humans from aspects of altruism that have evolved only in the human lineage. To date, no experimental studies have systemati￾cally tested human infants and chimpanzees in a similar set of helping situations. A number of studies have demonstrated that young children show concern (empathy) for others in distress. Preschool-age children and even infants (1 to 2 years of age) occasionally attempt to respond to the emotional needs of others, for example, by comforting someone who is crying (1–10). In contrast, there are no experimental studies with infants that have sys￾tematically investigated instrumental helping— providing help to people who are faced with an instrumental problem and are unable to reach their goal (11–13). In the current study we presented 24 18- month-old infants with 10 different situations in which an adult (a male experimenter) was having trouble achieving a goal. This variety of tasks presented the children with a variety of difficulties in discerning the adult_s goal and his problems in reaching the goal. These sit￾uations fell into four categories: out-of-reach objects, access thwarted by a physical obstacle, achieving a wrong (correctable) result, and using a wrong (correctable) means (Table 1) (movies S1 to S4). For each task, there was a corresponding control task in which the same basic situation was present but with no in￾dication that this was a problem for the adult (14). This ensured that the infant_s motivation was not just to reinstate the original situation or to have the adult repeat the action, but rather to actually help the adult with his problem. After the occurrence of the problem in each task (e.g., marker drops on floor), there were three phases: The experimenter focused on the object only (1 to 10 s), then alternated gaze between object and child (11 to 20 s), and in addition verbalized his prob￾lem while continuing to alternate gaze (e.g., BMy marker![; 21 to 30 s). In control trials, he looked at the object with a neutral facial expression for 20 s. In no case did the infant receive any benefit (reward or praise) for helping. Each individual was tested in all 10 tasks, a subsample of 5 tasks administered as experimental and 5 as control conditions (in systematically varied order). Thus, in each task 12 children were tested in the experimen￾tal condition and 12 others in the control con￾dition for a between-subjects comparison. Results showed that infants helped the adult (the infant performed the target behavior signif￾icantly more in experimental than in control conditions) in 6 of the 10 tasks—at least one for each category (Fig. 1). They handed him several out-of-reach objects (but not if he had discarded them deliberately); they completed his stacking of books after his failed attempt (but not if his placement of the books appeared to meet his goal); they opened the door of a cabinet for him when his hands were full (but not if he struggled toward the top of the cabinet); and they retrieved an inaccessible object for him by opening a box using a means he was unaware of (but not if he had thrown the object inside the box on purpose). Analyzed by individual, 22 of the 24 infants helped in at least one of the tasks. It is noteworthy that they did so in almost all cases immediately (average latency 0 5.2 s), before the adult either looked to them or verbalized his problem (84% of helping acts within the initial 10-s phase). Thus, the exper￾imenter never verbally asked for help, and for the vast majority of helping acts, eye contact (as a subtle means of soliciting help) was also unnecessary. Experimental studies on altruistic behaviors in nonhuman primates are scarce. There are anecdotal reports of possible instances of helping (15–17) and some experiments dem￾onstrating empathic intervention by various monkey species when another individual is displaying emotional distress (but no exper￾iments with apes) (18). However, there are no studies, to our knowledge, of nonhuman primates helping others who are struggling to achieve their goals (instrumental helping) (19, 20). In two recent experiments, chimpan￾zees were given the opportunity to deliver food to a conspecific (21, 22), but again that con￾specific was not trying to solve a problem in which the subject could help instrumentally Esee also (23)^. Results were negative. But it is possible that altruism would be more likely when it involves objects other than food, be￾cause chimpanzees often compete over food and the drive to acquire food for themselves might preclude their capacity to act on behalf of others. In the current study, therefore, we gave the same basic tasks of instrumental helping given to the infants, with some minor mod￾ifications, to three young chimpanzees (Pan troglodytes, one of humans_ two closest living relatives). These individuals were 36, 54, and Department of Developmental and Comparative Psycholo￾gy, Max Planck Institute for Evolutionary Anthropology, Deutscher Platz 6, 04103 Leipzig, Germany. *To whom correspondence should be addressed. E-mail: warneken@eva.mpg.de Table 1. Examples of problems used in child study. Category Task Problem Out-of-reach Marker The adult accidentally drops a marker on the floor and unsuccessfully reaches for it (experimental) or intentionally throws a marker on the floor (control). Physical obstacle Cabinet The adult wants to put magazines into a cabinet, but the doors are closed so that he bumps into it (experimental) versus bumping into the doors as he tries to lift the magazines onto the cabinet (control). Wrong result Book A book slips from a stack as the adult attempts to place it on top of the stack (experimental) or he places it next to the stack (control). Wrong means Flap A spoon drops through a hole and the adult unsuccessfully tries to grasp it through the small hole, ignorant of a flap on the side of the box (experimental). Alternatively, he throws the spoon in the box on purpose (control). REPORTS www.sciencemag.org SCIENCE VOL 311 3 MARCH 2006 1301