Using robots to understand social behaviour

A major challenge in studying social behaviour stems from the need to disentangle the behaviour of each individual from the resulting collective. One way to overcome this problem is to construct a model of the behaviour of an individual, and observe whether combining many such individuals leads to the predicted outcome. This can be achieved by using robots. In this review we discuss the strengths and weaknesses of such an approach for studies of social behaviour. We find that robots—whether studied in groups of simulated or physical robots, or used to infiltrate and manipulate groups of living organisms—have important advantages over conventional individual‐based models and have contributed greatly to the study of social behaviour. In particular, robots have increased our understanding of self‐organization and the evolution of cooperative behaviour and communication. However, the resulting findings have not had the desired impact on the biological community. We suggest reasons for why this may be the case, and how the benefits of using robots can be maximized in future research on social behaviour.     

Emotion Attribution to a Non-Humanoid Robot in Different Social Situations

In the last few years there was an increasing interest in building companion robots that interact in a socially acceptable way with humans. In order to interact in a meaningful way a robot has to convey intentionality and emotions of some sort in order to increase believability. We suggest that human-robot interaction should be considered as a specific form of inter-specific interaction and that human–animal interaction can provide a useful biological model for designing social robots. Dogs can provide a promising biological model since during the domestication process dogs were able to adapt to the human environment and to participate in complex social interactions. In this observational study we propose to design emotionally expressive behaviour of robots using the behaviour of dogs as inspiration and to test these dog-inspired robots with humans in inter-specific context. In two experiments (wizard-of-oz scenarios) we examined humans'' ability to recognize two basic and a secondary emotion expressed by a robot. In Experiment 1 we provided our companion robot with two kinds of emotional behaviour (“happiness” and “fear”), and studied whether people attribute the appropriate emotion to the robot, and interact with it accordingly. In Experiment 2 we investigated whether participants tend to attribute guilty behaviour to a robot in a relevant context by examining whether relying on the robot''s greeting behaviour human participants can detect if the robot transgressed a predetermined rule. Results of Experiment 1 showed that people readily attribute emotions to a social robot and interact with it in accordance with the expressed emotional behaviour. Results of Experiment 2 showed that people are able to recognize if the robot transgressed on the basis of its greeting behaviour. In summary, our findings showed that dog-inspired behaviour is a suitable medium for making people attribute emotional states to a non-humanoid robot.     

From fish to fashion: experimental and theoretical insights into the evolution of culture

Recent years have witnessed a re-evaluation of the cognitive capabilities of fishes, including with respect to social learning. Indeed, some of the best experimental evidence for animal traditions can be found in fishes. Laboratory experimental studies reveal that many fishes acquire dietary, food site and mating preferences, predator recognition and avoidance behaviour, and learn pathways, through copying other fishes. Concentrating on foraging behaviour, we will present the findings of laboratory experiments that reveal social learning, behavioural innovation, the diffusion of novel behaviour through populations and traditional use of food sites. Further studies reveal surprisingly complex social learning strategies deployed by sticklebacks. We will go on to place these observations of fish in a phylogenetic context, describing in which respects the learning and traditionality of fish are similar to, and differ from, that observed in other animals. We end by drawing on theoretical insights to suggest processes that may have played important roles in the evolution of the human cultural capability.     

Social behaviour of dogs encountering AIBO, an animal-like robot in a neutral and in a feeding situation

The use of animal-like autonomous robots might offer new possibilities in the study of animal interactions, if the subject recognises it as a social partner. In this paper we investigate whether AIBO, a dog-like robot of the Sony Corp. can be used for this purpose. Twenty-four adult and sixteen 4-5 months old pet dogs were tested in two situations where subjects encountered one of four different test-partners: (1) a remote controlled car; (2) an AIBO robot; (3) AIBO with a puppy-scented furry cover; and (4) a 2-month-old puppy. In the neutral situation the dog could interact freely with one of the partners for 1 min in a closed arena in the presence of its owner. In the feeding situation the encounters were started while the dog was eating food. Our results show that age and context influence the social behaviour of dogs. Further, we have found that although both age groups differentiated the living and non-living test-partners for some extent, the furry AIBO evoked significantly increased responses in comparison to the car. These experiments show the first steps towards the application of robots in behavioural studies, notwithstanding that at present AIBO's limited ability to move constrains its effectiveness as social partner for dogs.     

Evolving coordinated group behaviours through maximisation of mean mutual information

A well known problem in the design of the control system for a swarm of robots concerns the definition of suitable individual rules that result in the desired coordinated behaviour. A possible solution to this problem is given by the automatic synthesis of the individual controllers through evolutionary or learning processes. These processes offer the possibility to freely search the space of the possible solutions for a given task, under the guidance of a user-defined utility function. Nonetheless, there exist no general principles to follow in the definition of such a utility function in order to reward coordinated group behaviours. As a consequence, task dependent functions must be devised each time a new coordination problem is under study. In this paper, we propose the use of measures developed in Information Theory as task-independent, implicit utility functions. We present two experiments in which three robots are trained to produce generic coordinated behaviours. Each robot is provided with rich sensory and motor apparatus, which can be exploited to explore the environment and to communicate with other robots. We show how coordinated behaviours can be synthesised through a simple evolutionary process. The only criteria used to evaluate the performance of the robotic group is the estimate of mutual information between the motor states of the robots.     

Role of expressive behaviour for robots that learn from people

Robotics has traditionally focused on developing intelligent machines that can manipulate and interact with objects. The promise of personal robots, however, challenges researchers to develop socially intelligent robots that can collaborate with people to do things. In the future, robots are envisioned to assist people with a wide range of activities such as domestic chores, helping elders to live independently longer, serving a therapeutic role to help children with autism, assisting people undergoing physical rehabilitation and much more. Many of these activities shall require robots to learn new tasks, skills and individual preferences while ‘on the job’ from people with little expertise in the underlying technology. This paper identifies four key challenges in developing social robots that can learn from natural interpersonal interaction. The author highlights the important role that expressive behaviour plays in this process, drawing on examples from the past 8 years of her research group, the Personal Robots Group at the MIT Media Lab.     

Identifying Social Learning in Animal Populations: A New ‘Option-Bias’ Method

Background

Studies of natural animal populations reveal widespread evidence for the diffusion of novel behaviour patterns, and for intra- and inter-population variation in behaviour. However, claims that these are manifestations of animal ‘culture’ remain controversial because alternative explanations to social learning remain difficult to refute. This inability to identify social learning in social settings has also contributed to the failure to test evolutionary hypotheses concerning the social learning strategies that animals deploy.

Methodology/Principal Findings

We present a solution to this problem, in the form of a new means of identifying social learning in animal populations. The method is based on the well-established premise of social learning research, that - when ecological and genetic differences are accounted for - social learning will generate greater homogeneity in behaviour between animals than expected in its absence. Our procedure compares the observed level of homogeneity to a sampling distribution generated utilizing randomization and other procedures, allowing claims of social learning to be evaluated according to consensual standards. We illustrate the method on data from groups of monkeys provided with novel two-option extractive foraging tasks, demonstrating that social learning can indeed be distinguished from unlearned processes and asocial learning, and revealing that the monkeys only employed social learning for the more difficult tasks. The method is further validated against published datasets and through simulation, and exhibits higher statistical power than conventional inferential statistics.

Conclusions/Significance

The method is potentially a significant technological development, which could prove of considerable value in assessing the validity of claims for culturally transmitted behaviour in animal groups. It will also be of value in enabling investigation of the social learning strategies deployed in captive and natural animal populations.     

Socially intelligent robots: dimensions of human-robot interaction

Social intelligence in robots has a quite recent history in artificial intelligence and robotics. However, it has become increasingly apparent that social and interactive skills are necessary requirements in many application areas and contexts where robots need to interact and collaborate with other robots or humans. Research on human-robot interaction (HRI) poses many challenges regarding the nature of interactivity and 'social behaviour' in robot and humans. The first part of this paper addresses dimensions of HRI, discussing requirements on social skills for robots and introducing the conceptual space of HRI studies. In order to illustrate these concepts, two examples of HRI research are presented. First, research is surveyed which investigates the development of a cognitive robot companion. The aim of this work is to develop social rules for robot behaviour (a 'robotiquette') that is comfortable and acceptable to humans. Second, robots are discussed as possible educational or therapeutic toys for children with autism. The concept of interactive emergence in human-child interactions is highlighted. Different types of play among children are discussed in the light of their potential investigation in human-robot experiments. The paper concludes by examining different paradigms regarding 'social relationships' of robots and people interacting with them.     

Social learning in animals: comparative analysis of different forms and levels

Social learning as one of the key concepts of cognitive ecology includes different forms of behavioural displays from relatively simple, such as "social release" and "stimulus enhancement" up to "teaching" and "cultural transmission" in animal communities. Rapid development of this fields resulted in some contradictions in methods and terminology. In this review different forms and levels of social learning are analyzed. Ecological aspects of social learning are connected with diet shaping, fear of predators and mate choice. The first aspect is the most studied but still discussible. Social learning being an intricate component of feeding behaviour matches with innate behaviour, imprinting as well as early associative learning. Investigation of cognitive aspects of social learning going back to Thorndike's crucial question "Do apes ape?" are now developing into series of questions including even: "Do ants ape?". Elaboration of universal methods of comparative studying of social learning such as "artificial fruit" and "two ways/one outcome" has essentially enlightened these questions and made comparative analysis possible. Large continuum of displays of cognitive skills in social learning has been revealed in non-primate species. One of the discussible issues in the role of social learning is distribution of innovations. Many authors have investigated this intriguing aspect of animal behaviour in different ways, such as long field observations as well as laboratory experiments based on "artificial innovators" that is specimens specially taught by experimentalists. Many impressive results were obtained; in particular it turned out in contradiction with some mathematical models that individuals in groups are rather different in their psychophysiological predisposition to innovative behaviour. Role of teaching in such different forms of behaviour as shaping of species-specific behavioural patterns and spread of innovations is considered. Although the majority of animals in wild populations are not good teachers and pupils, some cultural aspects of behaviour were recently revealed, mostly in primates. At the same time some classical results concerning cultural transmission of new patterns (for example, bottle-opening in tits) were experimentally revised. Many problems still remain unsolved, in particular, how spread of innovations may favour prosperity of populations; to what degree behavioural peculiarities of local groups may be determined by processes of social learning; which internal and external factors and under what circumstances invest into social learning in natural environment.     

Why Robots Should Be Social: Enhancing Machine Learning through Social Human-Robot Interaction

Social learning is a powerful method for cultural propagation of knowledge and skills relying on a complex interplay of learning strategies, social ecology and the human propensity for both learning and tutoring. Social learning has the potential to be an equally potent learning strategy for artificial systems and robots in specific. However, given the complexity and unstructured nature of social learning, implementing social machine learning proves to be a challenging problem. We study one particular aspect of social machine learning: that of offering social cues during the learning interaction. Specifically, we study whether people are sensitive to social cues offered by a learning robot, in a similar way to children’s social bids for tutoring. We use a child-like social robot and a task in which the robot has to learn the meaning of words. For this a simple turn-based interaction is used, based on language games. Two conditions are tested: one in which the robot uses social means to invite a human teacher to provide information based on what the robot requires to fill gaps in its knowledge (i.e. expression of a learning preference); the other in which the robot does not provide social cues to communicate a learning preference. We observe that conveying a learning preference through the use of social cues results in better and faster learning by the robot. People also seem to form a “mental model” of the robot, tailoring the tutoring to the robot’s performance as opposed to using simply random teaching. In addition, the social learning shows a clear gender effect with female participants being responsive to the robot’s bids, while male teachers appear to be less receptive. This work shows how additional social cues in social machine learning can result in people offering better quality learning input to artificial systems, resulting in improved learning performance.     

Beyond existence and aiming outside the laboratory: estimating frequency-dependent and pay-off-biased social learning strategies

The existence of social learning has been confirmed in diverse taxa, from apes to guppies. In order to advance our understanding of the consequences of social transmission and evolution of behaviour, however, we require statistical tools that can distinguish among diverse social learning strategies. In this paper, we advance two main ideas. First, social learning is diverse, in the sense that individuals can take advantage of different kinds of information and combine them in different ways. Examining learning strategies for different information conditions illuminates the more detailed design of social learning. We construct and analyse an evolutionary model of diverse social learning heuristics, in order to generate predictions and illustrate the impact of design differences on an organism's fitness. Second, in order to eventually escape the laboratory and apply social learning models to natural behaviour, we require statistical methods that do not depend upon tight experimental control. Therefore, we examine strategic social learning in an experimental setting in which the social information itself is endogenous to the experimental group, as it is in natural settings. We develop statistical models for distinguishing among different strategic uses of social information. The experimental data strongly suggest that most participants employ a hierarchical strategy that uses both average observed pay-offs of options as well as frequency information, the same model predicted by our evolutionary analysis to dominate a wide range of conditions.     

The evolution of social learning rules: Payoff-biased and frequency-dependent biased transmission

Humans and other animals do not use social learning indiscriminately, rather, natural selection has favoured the evolution of social learning rules that make selective use of social learning to acquire relevant information in a changing environment. We present a gene-culture coevolutionary analysis of a small selection of such rules (unbiased social learning, payoff-biased social learning and frequency-dependent biased social learning, including conformism and anti-conformism) in a population of asocial learners where the environment is subject to a constant probability of change to a novel state. We define conditions under which each rule evolves to a genetically polymorphic equilibrium. We find that payoff-biased social learning may evolve under high levels of environmental variation if the fitness benefit associated with the acquired behaviour is either high or low but not of intermediate value. In contrast, both conformist and anti-conformist biases can become fixed when environment variation is low, whereupon the mean fitness in the population is higher than for a population of asocial learners. Our examination of the population dynamics reveals stable limit cycles under conformist and anti-conformist biases and some highly complex dynamics including chaos. Anti-conformists can out-compete conformists when conditions favour a low equilibrium frequency of the learned behaviour. We conclude that evolution, punctuated by the repeated successful invasion of different social learning rules, should continuously favour a reduction in the equilibrium frequency of asocial learning, and propose that, among competing social learning rules, the dominant rule will be the one that can persist with the lowest frequency of asocial learning.     

Distributed scalable multi-robot learning using particle swarm optimization

Designing effective behavioral controllers for mobile robots can be difficult and tedious; this process can be circumvented by using online learning techniques which allow robots to generate their own controllers online in an automated fashion. In multi-robot systems, robots operating in parallel can potentially learn at a much faster rate by sharing information amongst themselves. In this work, we use an adapted version of the Particle Swarm Optimization algorithm in order to accomplish distributed online robotic learning in groups of robots with access to only local information. The effectiveness of the learning technique on a benchmark task (generating high-performance obstacle avoidance behavior) is evaluated for robot groups of various sizes, with the maximum group size allowing each robot to individually contain and manage a single PSO particle. To increase the realism of the technique, different PSO neighborhoods based on limitations of real robotic communication are tested and compared in this scenario. We explore the effect of varying communication power for one of these communication-based PSO neighborhoods. To validate the effectiveness of these learning techniques, fully distributed online learning experiments are run using a group of 10 real robots, generating results which support the findings from our simulations.     

The context of social learning: association patterns in a captive flock of brown-headed cowbirds

Much work on social learning has involved behaviour transmission between pairs of individuals, but recently the need to examine the social context in which learning occurs has been recognized. Previous studies using small numbers of animals have shown social influence on the behavioural development of juvenile male brown-headed cowbirds, Molothrus ater. Here we looked at the larger social context that forms the framework for such influence in more natural settings. We allowed a captive group of over 70 cowbirds, comprising adult and juvenile males and females, to associate freely in a large complex of connected aviaries. Highly organized social assortment emerged in the group, with individuals associating with others based on similarity in age and sex. Juvenile males that associated more with adult males had higher courtship success. Juvenile males that associated more with females sang less over the year. These results indicate that the social context of social learning for juvenile males is not just random association with all other birds in the social group, but is a selective and structured pattern of interaction. Differences in navigating this social structure correlated with courtship success and vocalization, behaviour known to be affected by social learning. Studies such as this, using large groups with free assortment of individuals, are the first step towards understanding the effects of the larger social context surrounding social learning.     

Hominid cultural transmission and the evolution of language

This paper presents the hypothesis that linguistic capacity evolved through the action of natural selection as an instrument which increased the efficiency of the cultural transmission system of early hominids. We suggest that during the early stages of hominization, hominid social learning, based on indirect social learning mechanisms and true imitation, came to constitute cumulative cultural transmission based on true imitation and the approval or disapproval of the learned behaviour of offspring. A key factor for this transformation was the development of a conceptual capacity for categorizing learned behaviour in value terms - positive or negative, good or bad. We believe that some hominids developed this capacity for categorizing behaviour, and such an ability allowed them to approve or disapprove of their offsprings- learned behaviour. With such an ability, hominids were favoured, as they could transmit to their offspring all their behavioural experience about what can and cannot be done. This capacity triggered a cultural transmission system similar to the human one, though pre-linguistic. We suggest that the adaptive advantage provided by this new system of social learning generated a selection pressure in favour of the development of a linguistic capacity allowing children to better understand the new kind of evaluative information received from parents.     

Opening the Black Box of Developmental Experiments: Behavioural Mechanisms Underlying Long‐Term Effects of Early Social Experience

Typically, animals spend a considerable portion of their time with social interactions involving mates, offspring, competitors and group members. The social performance during these interactions can strongly depend on the social environment individuals have experienced early in life. Despite a considerable number of experiments investigating long‐term effects of the early social environment, our understanding of the behavioural mechanisms mediating these effects is still limited, mainly for two reasons. (1) Only in few experimental studies have researchers actually observed and quantified the behaviour of their study animals during the social treatment. (2) Even if differences in social interactions between social rearing treatments are reported, these differences might not be causally linked to any observed long‐term effects later in life. The aim of this review was to investigate whether behavioural records of animals during the experimental manipulation of their social environment can help (1) identifying behavioural mechanisms involved in a long‐term effect and (2) obtaining a better understanding of the long‐term consequences of early manipulations. First, I review studies that manipulated the social environment at an early stage of the ontogeny, observed the social interactions and behaviour during the social experience phase and subsequently tested the performance in social and non‐social behavioural tasks at a later life stage. In all reviewed studies, treatment differences were reported both in social interactions during the social experience phase and in social and/or non‐social behaviours later in life. Second, I discuss four classes of behavioural mechanisms that can cause the reported long‐term effects of social experience, namely learning by experience, social learning, sensory stimulation and social cueing. I conclude that social interactions during the social experience phase should always be recorded for at least two reasons. Knowledge about how the social interactions differ between rearing treatments (1) permits researchers to formulate hypotheses about candidate mechanisms causing long‐term effects on behaviour and (2) can help to interpret unexpected outcomes of developmental experiments. Finally, I propose that as a crucial ultimate step towards understanding effects of the early social environment, we should develop targeted experiments testing for the causality of identified candidate mechanism.     

Design, Development and Decisions

Evolutionary ideas and modern biological knowledge have important roles to play in the understanding of human behaviour. Nevertheless, it is deeply misleading to regard humans as robots in the grip of their genes. A well designed brain should respond to the consequences of behaviour; if an understanding of the likely consequences can be achieved without actually performing the act, then a person who knows that they will be rewarded or punished for certain acts is bound to be influenced by that knowledge. A brain designed in that way facilitates the evolution of societies with explicit social approval of certain activities and explicit disapproval of others. The evolutionary approach to psychology does not imply that individuals do not make free choices. Individuals clearly do make a big difference to what happens in their lives through their decisions. They may be surprised by the consequences of their own actions. A well designed brain should be able to anticipate the consequences of various courses of action and choose between them on the basis of their likely costs and benefits. Planning before doing is clearly of great advantage. People do make well considered decisions and they benefit from doing so.     

Social learning of novel route preferences in adult humans

Non-human animals can acquire novel route preferences by following knowledgeable individuals. Such socially learned route preferences can be stably maintained over multiple transmission episodes, sometimes forming long-lived traditions. In humans, preferences for familiar routes or heavily used worn trails over unfamiliar ones have been described in various contexts. However, social learning of route preferences has not been experimentally demonstrated in humans. Here, we demonstrate that social learning and tradition influence route choice. We led adult male and female participants into a room by one of two routes. Participants followed the demonstrated route choices, and later remembered and preferred this choice even when determinably suboptimal (i.e. longer and not preferred by control participants) or when the choice was indicated as arbitrary (the demonstrator took one route to retrieve a poster that had ostensibly fallen). Moreover, route preferences were stably maintained over multiple transmission episodes. We suggest that simple social learning processes, often neglected in human and primate research, can result in long-lived route preferences that may influence a range of additional behaviour patterns.     

Social learning in birds and its role in shaping a foraging niche

We briefly review the literature on social learning in birds, concluding that strong evidence exists mainly for predator recognition, song, mate choice and foraging. The mechanism of local enhancement may be more important than imitation for birds learning to forage, but the former mechanism may be sufficient for faithful transmission depending on the ecological circumstances. To date, most insights have been gained from birds in captivity. We present a study of social learning of foraging in two passerine birds in the wild, where we cross-fostered eggs between nests of blue tits, Cyanistes caeruleus and great tits, Parus major. Early learning causes a shift in the foraging sites used by the tits in the direction of the foster species. The shift in foraging niches was consistent across seasons, as showed by an analysis of prey items, and the effect lasted for life. The fact that young birds learn from their foster parents, and use this experience later when subsequently feeding their own offspring, suggests that foraging behaviour can be culturally transmitted over generations in the wild. It may therefore have both ecological and evolutionary consequences, some of which are discussed.     

Juvenile Galápagos Pelicans Increase Their Foraging Success by Copying Adult Behaviour

Social learning is the building block of culture and traditions in humans and nonhuman animals, and its study has a long history. Most investigations have addressed either the causation or the function of social learning. Though much is known about the underlying mechanisms of social learning, demonstrations of its adaptive value in a natural setting are lacking. Here we show that juvenile brown pelicans (Pelecanus occidentalis) can increase their foraging efficiency by copying adult diving behaviour, suggesting that social learning helps juveniles to find profitable food patches. Our findings demonstrate the potential fitness consequences of behavioural copying and thus highlight the possible adaptive importance of social learning.