- Development microeconomics;
- An introduction to probability theory.
- Multi-Level Governance in Universities: Strategy, Structure, Control.
- Account Options;
For each randomisation, we randomly allocated individuals to groups, constraining group sizes to remain the same as that observed, and calculated the resulting option bias statistic. We repeated this procedure 10, times to generate a null distribution, and calculated the p-value as the proportion of the null distribution that was greater than or equal to the observed option bias statistic. Since the data are randomised at the level of individuals, the test allows for the fact that manipulations by the same individual might not be independent.
The Supplementary Material, S1. Randomisation is an established technique for testing hypotheses when the assumptions of conventional statistical techniques are not met . Nonetheless, we wanted to validate the method in the context of detecting social learning in groups of animals. We did this using simulations of the asocial and social learning process for each task, enabling us to assess type 1 error rate and statistical power, and compare the performance of the method to other candidate techniques.
These were a Fisher's Exact Test, run on data reduced to a single datum for each individual, representing its mean response; b a GLMM with a binomial error structure, testing for a fixed effect of group using a likeihood ratio test LRT with individual as a random effect; c randomisation techniques using a test statistic generated from i a GLM with a binomial error structure and ii a log linear model, both with group as an explanatory variable; and d generation of a null distribution through a Monte Carlo simulation of the asocial learning process.
The simulations also allowed us to estimate the power of the chosen technique to detect a given effect size for each of the tasks see Figures 2 and 3. Full details of the simulations can be found in the Supplementary Material S1. For details see text. Dotted lines show Wilson's confidence intervals. In all cases the power is higher than that for Fisher's exact test.
See supplementary material S1. Traditional laboratory studies infer social learning by showing that an observer is more likely to solve a task using the same option as its demonstrator used compared with an alternative. If the option bias method is a suitable technique for detecting social learning in groups of animals, we would expect its findings to match those of the traditional approach when applied to the same data.
Studies by Coolen et al. In Coolen et al. Therefore, we would expect to find a significant option bias for 9-spined sticklebacks, but not for 3-spined sticklebacks. Coolen et al. The fish were then assigned to groups on the basis of the demonstration they had observed: a rich right patch or a rich left patch. Social learning is then inferred if the individuals in each group tend to adopt the method of their demonstrator. Whiten et al. An analysis of the callitrichid data confirmed the existence of significant differences in the difficulty of the three tasks, as gauged by both latency to solve and number of unsuccessful task manipulations and successful extractions by the monkeys.
The measure of learning time controls for time to first contact, and so these differences cannot be attributed to variation in task salience or neophobia to the task. These findings indicate that the cylinder was the most difficult, and the round-box the simplest, task to solve. The estimated power of these results is shown in Figure 3 and indicates that there is insufficient power to determine whether or not there is evidence for social learning with the cylinder task.
Of the innovators, approximately equal numbers first used each task option flip-top: 6 blue vs. We therefore consider this an inappropriate method for analysing option bias. The estimated power for the Monte Carlo simulation was always lower than that of the randomisation methods. All randomisation methods had better power than Fisher's exact test, primarily because they take into account the strength of individual-level option bias, rather than just its direction.
The relative performance of these tests is shown in Fig. The performance of the randomisation techniques was very similar, consequently we consider any of these approaches appropriate for analysing option bias data. The estimated power for this method is shown in Fig. In agreement with Coolen et al. Likewise, application of the option bias technique to Whiten et al. That the option bias is able to detect social learning in this more naturalistic context supports our claim that it will be able to detect social learning of naturally occurring behavioural variants.
We have introduced a new method, called the option-bias method , for detecting social learning in animal populations. The method is based on a widely applied premise in social learning research, the assumption that - when ecological and genetic differences are accounted for - social learning will generate greater homogeneity in behaviour within groups than expected in its absence. Our analyses reveal that the approach gives greater power to detect social learning than conventional inferential statistics, while its validity is confirmed through application to established experimental datasets.
We have applied the method to experimental data collected from groups of callitrichid monkeys provided with three novel foraging tasks. We conclude that there is compelling evidence for social learning in only one of these tasks, namely the flip-top task. It is highly unlikely that the observed option bias for successful manipulations in this task arose by chance under asocial individual learning alone.
Conversely, given the relatively high power estimated for the round-box task Figure 3 , it seems likely that there is no social learning, or its effects are very weak for this task. Since the round-box was the easiest task, this finding is consistent with the hypothesis that social learning will only be used for difficult tasks .
Whilst the cylinder was found to be the most difficult task, the estimated power for this task was much lower over a plausible range of values for the strength of social learning Figure 3. Here the low power reflects the small number of populations for which data is available, and small number of successful manipulations, rather than any intrinsic feature of the option-bias method.
Therefore, we cannot rule out a meaningful role for social learning for the cylinder task, and think it likely that further data collection would demonstrate this. These results are broadly consistent with the predictions by Boyd and Richerson  ,  and other researchers [e. Here the best evidence for social learning is found in the flip-top task, while we cannot rule out social learning in the case of the cylinder; these are the two most difficult tasks.
Do Animals Have Memes?
Conversely, the solution to the round box task, demonstrated to be easiest, is almost certainly acquired asocially. Plausibly, the monkeys may have adopted social learning to avoid the cost of excess expenditure of time or energy and reduced foraging success associated with asocial learning of complex tasks.
The relatively high proportion of unsuccessful manipulations produced by innovators tackling the cylinder compared with those tackling the round-box implies that more trial-and-error-learning or discovery is involved in achieving success with the cylinder than round-box task. Where trial-and-error asocial learning is minimal i. This interpretation is supported by a suite of recent experiments involving two-action tasks with chimpanzees that point to a positive relationship between reliance on social learning and task difficulty .
The observed significant bias for use of one option over the other with the flip-top task implies that the actions of demonstrating individuals drew the attention of conspecifics to a specific option, causing them to direct the majority of their manipulations towards it. The converse finding, that social learning is not involved in generating option biases in unsuccessful manipulations of the tasks, is consistent with the idea that animals may distinguish between functionally relevant and irrelevant information, or rewarded and unrewarded information, and preferentially learn about the former [e.
This interpretation implies localised stimulus enhancement [ sensu 35] in combination with an emulative, or goal directed, process  or observational conditioning [ sensu 37]. Alternatively, there may be an exploratory phase early on, corresponding to an individual monkey's unsuccessful manipulations, prior to their option choices being biased by the successful behaviour of conspecifics. Our option-bias method could potentially be widely applied within the field of social learning and culture. Thus the method may prove useful to other researchers attempting to distinguish social and asocial learning in social contexts and provides a new and potentially valuable tool for the identification of cultural traditions.
The method could be deployed within controlled experimental and captive animal settings and, with the below caveats, to natural datasets too. However, it is important to emphasise the underlying assumption of the method — social learning leads to homogeneity of behaviour — and the consequent need for researchers to account for other factors eg.
For example, where a group is very large, or the mode of transmission is believed to be purely vertical, heterogeneity of behaviour may be expected between cliques, or genetically related sub-groups, necessitating analysis at this level. The method is applicable to instances where tasks have any number of options available for solution. Data on the presence or absence of a potential socially learned preference i.
The method could also be applied to natural behaviour for which there exist two or more variants. Naturally occurring phenomena that might fit this context include chimpanzee termite fishing where either one end or both ends of the tool e. The fact that the method does not require the researcher to record the inception and initial spread of the trait further enhances its utility in natural populations. Importantly, the method can legitimately be applied in cases where there is an unequal prior probability of performing the two options, for instance, as a result of ecological or genetic variation; indeed, we find that the relative power of our method is enhanced in such cases compared to Fisher's exact test see Supplementary material, S1.
The method is not designed to distinguish unlearned from learned behaviour, and would have to be employed in conjunction with other approaches to partial out any influence of genetic variation on population-level differences in behaviour [e. However the presence of genetic or ecological differences between populations in the probability of option use does not a priori rule out application of the method.
Our analyses reveal that provided independent estimates of the probability of option use can be generated for each population, the method will generate reliable results. In cases where such estimates are not available, researchers are forced to apply the method on a smaller scale, at which such variation does not apply. We note here that the Monte Carlo simulation variant of the approach can be applied to detect social learning in a single population see further explanation below.
While the option-bias approach is not foolproof, and like any statistical approach is vulnerable to both Type I and Type II errors, our analyses reveal that it is associated with greater statistical power and lower error than alternative methods. While here we have placed emphasis on a variant of the method that utilizes randomization to compute asocial probabilities, we do not rule out the possibility that other methods for generating the asocial sampling distribution could prove useful in other circumstances.
For instance, the Monte Carlo method described in the supplementary material S1 has the added advantage that it can be used to detect a significant option bias in a single group of animals. Here, only one statistic e. The increasing realization among behaviorists and psychologists is that many animals learn by observation as members of social systems. Such settings contribute to the formation of culture. This book combines the knowledge of two groups of scientists with different backgrounds to establish a working consensus for future research.
The book is divided into two major sections, with contributions by a well-known, international, and interdisciplinary team which integrates these growing areas of inquiry. Integrates the broad range of scientific approaches being used in the studies of social learning and imitation, and society and culture Provides an introduction to this field of study as well as a starting point for the more experienced researcher Chapters are succinct reviews of innovative discoveries and progress made during the past decade Includes statements of varied theoretical perspectives on controversial topics Authoritative contributions by an international team of leading researchers.
Though the idea of 'culture' in animals has only been around for just over half of a century, scientists have been noting social behaviors of animals for centuries. Aristotle was the first to provide evidence of social learning in the songs of birds. Darwin was also the first to suggest what became known as social learning in attempting to explain the transmission of an adaptive pattern of behavior through a population of honey bees. The vast majority of cultural anthropological research has been done on non-human primates, due to their being closest evolutionarily to humans. In non-primate animals, research tends to be limited, and therefore evidence for culture strongly lacking.
However, the subject has become more popular recently, and has prompted the initiation of more research into the area. Evolutionary biologist Richard Dawkins made groundbreaking headway into the field of cultural transmission with his book entitled The Selfish Gene , which focused heavily on the move to evolution being understood primarily by genetic influence.
Dawkins coined the term meme , the primary unit of cultural transmission or imitation, to explain an overarching mechanism of how animal behavior is shared and spread to lead to cultural evolution. The use of the word meme was an intentional phonetic derivation of the similar sounding word " gene ", which Dawkins asserts to be the primary unit of selection as it lends itself to pathways of biological evolution:. I hope my classicist friends will forgive me if I abbreviate mimeme to meme.
However, with the meme unit, cultural transmission has a distinct feature of being capable of taking place by individuals developing varying interpretations of the meme without exactly "copying" it to pass it on. These interpretations lead to the creation of new memes, which are themselves subject to a cyclic process of selection, rejection, or modification. Andrews , contributed to the greater understanding of cultural transmission with his work on chimpanzees.
Social Learning In Animals
In Cultural Traditions in Chimpanzees , Whiten created a compilation of results from seven long-term studies totaling years of observation analyzing behavioral patterns in different communities of chimpanzees in Africa read more about it below. The study expanded the notion that cultural behavior lies beyond linguistic mediation, and can be interpreted to include distinctive socially learned behavior such as stone-handling and sweet potato washing in Japanese macaques.
Cultural transmission , also known as cultural learning , is the process and method of passing on socially learned information. Differences in cultural transmission across species have been thought to be largely affected by external factors, such as the physical environment, that may lead an individual to interpret a traditional concept in a novel way. The environmental stimuli that contribute to this variance can include climate , migration patterns , conflict, suitability for survival, and endemic pathogens.
Cultural transmission can also vary according to different social learning strategies employed at the species and or individual level. Culture , when defined as the transmission of behaviors from one generation to the next, can be transmitted among animals through various methods. Imitation has been found to be one of the most prevalent modes of cultural transmission in non-human animals, while teaching and language are much less widespread, with the possible exceptions of primates and cetaceans.
Recent research has suggested that teaching, as opposed to imitation, may be a characteristic of certain animals who have more advanced cultural capacities, though this is debatable. The likelihood of larger groups within a species developing and sharing these intra-species traditions with peers and offspring is much higher than that of one individual spreading some aspect of animal behavior to one or more members.
This is why cultural transmission has been shown to be superior to individual learning, as it is a more efficient manner of spreading traditions and allowing members of a species to collectively inherit more adaptive behavior. The role of cultural transmission in cultural evolution, then, is to provide the outlet for which organisms create and spread traditions that shape patterns of animal behavior visibly over generations. Culture, which was once thought of as a uniquely human trait, is now firmly established as a common trait among animals and is not merely a set of related behaviors passed on by genetic transmission as some have argued.
Genetic transmission, like cultural transmission, is a means of passing behavioral traits from one individual to another. The main difference is that genetic transmission is the transfer of behavioral traits from one individual to another through genes which are transferred to an organism from its parents during the fertilization of the egg. As can be seen, genetic transmission can only occur once during the lifetime of an organism. In cultural transmission, behavioral information is passed through means of verbal, visual, or written methods of teaching. Therefore, in cultural transmission, new behaviors can be learned by many organisms in a matter of days and hours rather than the many years of reproduction it would take for a behavior to spread among organisms in genetic transmission.
Culture can be transmitted among animals through various methods, the most common of which include imitation , teaching , and language. Imitation is one of the most prevalent modes of cultural transmission in non-human animals, while teaching and language are much less widespread. In a study  on food acquisition techniques in meerkats Suricata suricatta , researchers found evidence that meerkats learned foraging tricks through imitation of conspecifics. The experimental setup consisted of an apparatus containing food with two possible methods that could be used to obtain the food.
Teaching is often considered one mechanism of social learning,  and occurs when knowledgeable individuals of some species have been known to teach others. Until recently, teaching was a skill that was thought to be uniquely human. Teaching is not merely limited to mammals either. Many insects, for example have been observed demonstrating various forms of teaching in order to obtain food. Ants, for example, will guide each other to food sources through a process called " tandem running ", in which an ant will guide a companion ant to a source of food.
Killer whales are known to "intentionally beach" themselves in order to catch and eat pinnipeds who are breeding on the shore. Teaching is arguably the social learning mechanism that affords the highest fidelity of information transfer between individuals and generations, and allows a direct pathway through which local traditions can be passed down and transmitted. Imitation is often misinterpreted as merely the observation and copying of another's actions.
This would be known as mimicry, because the repetition of the observed action is done for no other purpose than to copy the original doer or speaker. In the scientific community, imitation is rather the process in which an organism purposefully observes and copies the methods of another in order to achieve a tangible goal. Recent research into imitation in animals has resulted in the tentative labeling of certain species of birds, monkeys, apes, and cetaceans as having the capacity for imitation.
For example, a Grey parrot by the name of Alex underwent a series of tests and experiments at the University of Arizona in which scientist Irene Pepperberg judged his ability to imitate the human language in order to create vocalizations and object labels.
Through the efforts of Pepperberg, Alex has been able to learn a large vocabulary of English words and phrases. Alex can then combine these words and phrases to make completely new words which are meaningless, but utilize the phonetic rules of the English language. Language is another key indicator of animals who have greater potential to possess culture.
Though animals do not naturally use words like humans when they are communicating, the well-known parrot Alex demonstrated that even animals with small brains, but are adept at imitation can have a deeper understanding of language after lengthy training. A bonobo named Kanzi has taken the use of the English language even further. Kanzi was taught to recognize words and their associations by using a lexigram board. Through observation of its mother's language training, Kanzi was able to learn how to use the lexigrams to obtain food and other items that he desired.
The beginning of the modern era of animal culture research in the middle of the 20th century came with the gradual acceptance of the term "culture" in referring to animals.
Japan's leading primatologist of the time, Kinji Imanishi , first used the word with a prefix as the term "pre-culture" in referring to the now famous potato-washing behavior of Japanese macaques. In , Imanishi and his colleagues began studying macaques across Japan, and began to notice differences among the different groups of primates, both in social patterns and feeding behavior.
One of the groups commonly dug up and ate the tubers and bulbs of several plants, while monkeys from other groups would not even put these in their mouths. Imanishi had reasoned that, "if one defines culture as learned by offspring from parents, then differences in the way of life of members of the same species belonging to different social groups could be attributed to culture. The most famous of these eating behaviors was observed on the island of Koshima , where one young female was observed carrying soiled sweet potatoes to a small stream, where she proceeded to wash off all of the sand and dirt before eating.
This behavior was then observed in one of the monkey's playmates, then her mother and a few other playmates. The potato-washing eventually spread throughout the whole macaque colony, encouraging Imanishi to refer to the behavior as "pre-culture," explaining that, "we must not overestimate the situation and say that 'monkeys have culture' and then confuse it with human culture. The first evidence of apparently arbitrary traditions came in the lates, also in the behavior of primates.
At this time, researchers McGrew and Tutin found a social grooming handclasp behavior to be prevalent in a certain troop of chimpanzees in Tanzania, but not found in other groups nearby. Though this would seem to make grooming of the armpits easier, the behavior actually has no apparent advantage.
As the primatologist Frans de Waal explains from his later observations of the hand-clasp grooming behavior in a different group of chimpanzees, "A unique property of the handclasp grooming posture is that it is not required for grooming the armpit of another individual Thus it appears to yield no obvious benefits or rewards to the groomers.
Prior to these findings, opponents to the idea of animal culture had argued that the behaviors being called cultural were simply behaviors that had evolutionarily evolved due to their importance to survival. After the identification of this initial non-evolutionarily advantageous evidence of culture, scientists began to find differences in group behaviors or traditions in various groups of primates, specifically in Africa. More than 40 different populations of wild chimpanzees have been studied across Africa, between which many species-specific, as well as population-specific, behaviors have been observed.
The researching scientists found 65 different categories of behaviors among these various groups of chimpanzees, including the use of leaves, sticks, branches, and stones for communication, play, food gathering or eating, and comfort. In , Whiten et al. The synthesis of their studies consisted of two phases, in which they 1 created a comprehensive list of cultural variant behavior specific to certain populations of chimpanzees and 2 rated the behavior as either customary — occurring in all individuals within that population; habitual — not present in all individuals, but repeated in several individuals; present — neither customary or habitual but clearly identified; absent — instance of behavior not recorded and has no ecological explanation; ecological — absence of behavior can be attributed to ecological features or lack thereof in the environment, or of unknown origin.
Their results were extensive: of the 65 categories of behavior studied, 39 including grooming , tool usage and courtship behaviors were found to be habitual in some communities but nonexistent in others. Whiten et al. This finding confirms not only that nonhuman species can maintain unique cultural traditions; it also shows that they can pass these traditions on from one population to another. The Whiten articles are a tribute to the unique inventiveness of wild chimpanzees, and help prove that humans' impressive capacity for culture and cultural transmission dates back to the now-extinct common ancestor we share with chimpanzees.
Similar to humans, social structure plays an important role in cultural transmission in chimpanzees. Victoria Horner conducted an experiment where an older, higher ranking individual and a younger, lower ranking individual were both taught the same task with only slight aesthetic modification.
It is believed that the older higher ranking individual had gained a level of 'prestige' within the group. This research demonstrates that culturally transmitted behaviors are often learned from individuals that are respected by the group. The older, higher ranking individual's success in similar situations in the past led the other individuals to believe that their fitness would be greater by imitating the actions of the successful individual.
This shows that not only are chimpanzees imitating behaviors of other individuals, they are choosing which individuals they should imitate in order to increase their own fitness. This type of behavior is very common in human culture as well.
People will seek to imitate the behaviors of an individual that has earned respect through their actions. From this information, it is evident that the cultural transmission system of chimpanzees is more complex than previous research would indicate. Chimpanzees have been known to use tools for as long as they have been studied. Andrew Whiten found that chimpanzees not only use tools, but also conform to using the same method as the majority of individuals in the group.
The results from the research of Victoria Horner and Andrew Whiten show that chimpanzee social structures and human social structures have more similarities than previously thought. Second only to non-human primates, culture in species within the order Cetacea , which includes whales , dolphins , and porpoises , has been studied for numerous years.
Social learning in animals : the roots of culture
In these animals, much of the evidence for culture comes from vocalizations and feeding behaviors. Cetacean vocalizations have been studied for many years, specifically those of the bottlenose dolphin, humpback whale, killer whale, and sperm whale. Hal Whitehead , a leading cetologist , and his colleagues conducted a study in of sperm whale groups in the South Pacific, finding that groups tended to be clustered based on their vocal dialects.
In mammals such as these sperm whales or bottlenose dolphins, the decision on whether an animal has the capacity for culture comes from more than simple behavioral observations. As described by ecologist Brooke Sergeant, "on the basis of life-history characteristics, social patterns, and ecological environments, bottlenose dolphins have been considered likely candidates for socially learned and cultural behaviors," due to being large-brained and capable of vocal and motor imitation.