What kind of social tendencies and cognitive evaluations urge an individual to help another or to cooperate on a joint task? In order to cooperate with others, individuals need to choose partners or join an ongoing effort. They may also be recruited by others, and will need to decide if the joint effort will be worth their participation. Sometimes they just assist others who are in trouble or cannot reach their goal alone. There are basically two situations:
(a) Anonymous cooperation: Species without individual recognition and remembered experiences cooperate with other conspecifics. Here partner choice is guided by cues that simply indicate that the other individual is a member of the same nest, colony, or population. Anonymous cooperation with members of the same group likely evolved under conditions in which all group members were closely related kin, and thus it was unnecessary (and perhaps too costly for large groups) to distinguish among them. Alternatively, if groups are fairly stable and interactions are repeated, cooperative behaviors may benefit all members of the group and thus be selected for, even if relatedness among individuals is low. Finally, social interactions (and the underlying neurophysiological and cognitive pathways that drive these interactions) that evolved among members of a closely related kin group may have easily expanded to include less related individuals. For example, phylogenetic studies have indicated that social behavior evolved under conditions in which groups were headed by singly mated females, and mat-
ing number increased subsequently. Thus, once social interactions among closely related individuals became firmly established, the group could take advantage of the benefits of increasing genetic diversity and thereby enhancing immunity, task specialization, and division of labor. Similarly, it is theorized that maternal care evolved into sib-care; in this case, the underlying sensory pathways simply expanded to include siblings as well as offspring. This is not unlike ideas about mammalian empathy (below), which is thought to be rooted in maternal care neural circuitry.
(b) Individualized cooperation: Species in which individuals recognize each other and build up a history of interaction known as “social relationships.”
Here we are concerned with individualized societies (e.g., mammals, birds, other vertebrates; de Waal & Tyack, 2003; also some invertebrates with individual recognition, such as paper wasps). Recruitment by means of communication is well developed, such as by vocalizations, postures, or gestures aimed at specific partners to solicit their support. For example, many primates have specific behaviors to activate supporters to help them in a fight. The decisions to cooperate are guided by previous experiences with the other (e.g., reliability and effectiveness) and its social rank, with high-ranking individuals being superior supporters. On the other hand, a partner of similar low rank is preferred to jointly overthrow the existing hierarchy. Choices are often guided by kinship, but also by mutual benefit and reciprocity, such as when chimpanzees share food with those who have groomed them before, or when fish recruit fish of another species that complements their hunting strategy.
Many of these cooperative situations are mutualistic; i.e., parties gain benefits at the same time. For example, when animals hunt in groups and together bring down large prey they benefit simultaneously. This kind of cooperation requires coordination, communication, and sharing of payoffs, but no altruism or long-term memory except memory of specific individuals and their effectiveness as partners. It also includes monitoring against freeloading.
The second kind of cooperation spreads benefits out over time so that one individual may gain now and his or her partner gains next time, an exchange known as reciprocity (Trivers, 1971). Several proximate mechanisms may produce reciprocity, but our understanding of how it works, and especially how it works outside of the primates, is still very limited. It is unlikely that animal actors know about reciprocity in the sense that they perform
helping acts with future return benefits in mind. There are indications that some large-brained species may do so, but the planning of future exchanges is beyond most animals’ cognition. This means that helping must have its own autonomous motivation. In humans, the main motivator is assumed to be empathy, which makes one person gain a stake in the other’s situation. Without necessarily implying the cognitively advanced forms found in human adults (e.g., theory of mind), the empathy explanation takes as its basis bodily connections, involuntary mimicry, and emotional contagion. These mechanisms have been proposed to underlie helping in other mammals as well, and there are indeed reports on empathy from mice to elephants, and also birds (de Waal, 2008).
Apart from kinship, what determines partner choice in a cooperative context? Is it (a) the effectiveness of the partner on the task at hand, (b) trust based on other experiences with the partner (e.g., affiliation and friendship), (c) the payoff division to be expected (i.e., tolerance, sharing), or (d) fear of retaliation (i.e., punishment for not cooperating)?
Apart from kinship, what determines the tendency to help others in distress or trouble? Empathy is generally thought to be promoted by similarity and familiarity between individuals.
What are the mechanisms underlying reciprocal exchange of favors? Is reciprocity organized along tit-for-tat lines, requiring memory and score-keeping, or are there simpler mechanisms at work, including the generalized reciprocity reported for a few species?
How similar are the proximate mechanisms mediating anonymous and individualized cooperation? Are the differences simply due to differences in the level of resolution of individual recognition?
de Waal FBM. Putting the altruism back into altruism: The evolution of empathy. Annual Review of Psychology 2008;59:279-300.
de Waal FBM and Tyack PL. Animal Social Complexity: Intelligence, Culture, and Individualized Societies. Harvard University Press, Cambridge, MA, 2003.
Linksvayer TA and Wade MJ. The evolutionary origin and elaboration of sociality in the aculeate Hymenoptera: Maternal effects, sib-social effects, and heterochrony. Quarterly Review of Biology 2005;80:317-336.
Trivers RL. The evolution of reciprocal altruism. Quarterly Review of Biology 1971;46:35-57.
IDR TEAM MEMBERS
Shawn Alter, Emory University
Sarah F. Brosnan, Georgia State University
Tyrone W.A. Grandison, Proficiency Labs
Mark E. Hauber, Hunter College, City University of New York
Katharine M. Jack, Tulane University
Timothy A. Linksvayer, University of Pennsylvania
Julie R. Murrell, EMD Millipore
Stacey R. Tecot, University of Arizona
IDR TEAM SUMMARY—GROUP 3
Shawn Alter, NAKFI Science Writing Scholar Emory University
IDR Team 3 was asked to determine the proximate mechanisms that give rise to cooperation between individuals. Cooperation is a core component of social evolution, allowing groups to adapt to selective pressures, even though it comes at a cost to altruistic individuals. This phenomenon is evident in all levels of life, from quorum-sensing unicellular bacteria to group-living primates. In 1964, William Hamilton proposed the theory of inclusive fitness, which holds that genetically similar individuals who cooperate indirectly improve the fitness of their group and the likelihood of passing on their shared, identical genes, including those that encode cooperative traits. However, inherited traits are ultimately the result of cooperation, rather than the proximate cause.
Primate ethologist and NAKFI Steering Committee member Frans de Waal challenged IDR Team 3 to find the very spark that ignites cooperation. This challenge is not just an exercise in scholarship, but may also provide fresh insights for the fields of behavioral economics, education, and artificial intelligence. Comprised of thought leaders from cellular through evolutionary biology, psychology, and computer science, the members of IDR Team 3 have collectively studied the social behaviors of ants, birds, primates, and simulated populations. By drawing from their collective expertise, IDR Team 3 established the conditions that are required for cooperation and compel organisms to cooperate and determined the proximate elements of cooperation. To determine the relative importance of these proximate elements—the cooperative spark—IDR Team 3 next turned to
the field of evolutionary robotics, and conceived the Artificial Test-bed for Experimentation into Cooperation and Helping (ATECH).
Evolutionary Robotics: A Very Unnatural Selection
The field of evolutionary robotics grew from a seed first planted by Alan Turing in the 1950s. In Computing machinery and Intelligence, Turing presaged that machines capable of learning and adaptation could not be created by man, but would rather be borne from an evolutionary process subject to mutation and selective reproduction. In the decades to follow, systems biologists developed computational algorithms to model and test evolutionary phenomena as the fields of artificial intelligence and robotics continued to flourish. In the 1990s, international teams of roboticists launched the first experiments in evolutionary robotics.
In an evolutionary robotics arena, robotic genomes are subject to survival of the fittest. In one of the earliest demonstrations, Dario Floreano, a pioneer in evolutionary robotics, challenged contenders to navigate a maze with mutably coded genomes that defined the patterns and activity of their neural networks; these networks in turn dictated the robots’ actions in response to sensory input during the challenge. Robots earned a fitness score (f), commensurate with their performance in the maze, so that robots that successfully navigated the maze with few crashes received high f scores, while those that crashed frequently or did not complete the maze received low f scores. Following the challenge, a computer would select the genomes of robots with the highest scores, and pair them for genetic crossover and mutation. The resultant genomes were then used to reprogram the robots as a new generation, in a process that would be reiterated hundreds of times.
Robots quickly evolved beyond maze navigation. In a chess-like battle, miniature predator and prey robots co-evolved hunting and evasion tactics, while robots capable of autonomous design and fabrication adapted brain and structural morphologies to adapt to varied mechanical tasks. In 2006, Floreano joined forces with evolutionary entomologist Laurent Keller to challenge multiple kindred of genetically similar robots to collectively forage and transport food tokens. These robots could boost their individual fitness by foraging and transporting small tokens by themselves, or opt to incur a fitness cost by foregoing this low-hanging fruit and cooperatively moving larger, otherwise immoveable tokens to earn a larger distributed fitness score. Remarkably, Keller and Floreano found that cooperative token movement evolved first and most strongly between the most genetically
similar robots. Hamilton’s theory of inclusive fitness thus applies to the field of evolutionary robotics, which may provide the ideal means to investigate the proximate mechanisms of cooperation.
Finding the Spark
Before ATECH could be formed, IDR Team 3 established the relevant testing parameters that would give rise to cooperation. IDR Team 3 defined cooperation as “interaction that, on average, benefits all participants,” which will be manifest in individual robots’ fitness scores. In cooperative interaction, continual iterations will produce a benefit greater than the participants could achieve individually, as is the case with collective carrying and food sharing among ants. In simulations and robotics, successful cooperation earns a collective fitness score that is distributed among the group. However, sometimes cheaters come along, draw from the fruits of cooperation, and do not expend any cost of their own, as is the case of a worker ant that forgoes food gathering to attempt mating and egg-laying to the detriment of the colony. Similarly, a “cheating” robot would accumulate group fitness points while leaving its kin to do all of the work; these groups are at risk of losing fitness to competing robot groups. IDR Team 3 hypothesized that the proximate elements below will provide a graded and interactive context for cooperation, whose relative strengths and effects can be directly tested.
Environmental variability. Within homogenous or genetically similar populations, environmental variability places selective pressure and drives specialization of individuals. Within a heterogeneous group in which specialization may already exist, environmental variability may exert demands that surpass the capacity of the individual participants.
Flexible response to cooperation. Faced with a challenge such as predation or limited resources, participants must be able to respond in such a manner as to serve their own interest or help other participants at a cost. Partners must be able to respond to each other’s actions and vary their own actions so that the sum effort is not so costly that it precludes collective benefit. In the case of altruism, however, an individual may incur great cost and even death to benefit the group, as is the case of a stinging honeybee that will die shortly after defending its queen and nest.
Communication. In order for participants to cooperate, it is essential that they be able to sense an environmental cue and recognize the cooperative “effort” of the other participants. For example, ant societies, which
have evolved the ability to divide labor and collectively solve problems, communicate using pheromones, sound, and touch.
Goal-directed activity. In order for participants to cooperate, they must share a common goal or interact in a manner that is beneficial to achieving each other’s goals, even though this goal need not be conscious. Goal-directed behavior occurs when one expends energy or effort toward an end, after which effort ceases (i.e., a feedback loop). Pack hunting by gray wolves is a prime example of this proximate element of cooperation.
Moving forward, IDR Team 3 proposes to manipulate the proximate elements of cooperation as test parameters to determine their impact on simulated and robotic fitness in cooperative challenges including foraging and defense from variably sized and challenging intruders. As ATECH develops and evolves, IDR Team 3 intends to develop predictive models that can be tested on model organisms to predict cooperation in nature.
This page intentionally left blank.