W ild chimpanzees produce 2 basic types of vocal alarms:

1. Crockford, Wittig, Mundry & Zuberbühler: Wild Chimpanzees Inform Ignorant Group Members of Danger. Current Biology 2012, 142-146.

2. Crockford, Wittig, Mundry & Zuberbühler: Wild Chimpanzees Inform Ignorant Group Members of Danger. Current Biology 2012, 142-146. • Chimps can understand some of the perceptual or motivational states of others, specifically what they see or intend to do (e.g., Hare, Call & Tomasello 2001; Kaminiski, Call & Tomasello 2010) • Are they simply reading the behavior of the other animal or attributing knowledgeto them (“a cognitively more complex process”)? • Previous investigations have typically been based on food paradigms, whereby another's ability to see food was equated with its intention to acquire it, “which has made it difficult to disentangle intention and perception from knowledge” • Alarm-calling behavior provides a useful alternative way to investigate the ability of primates to attribute knowledge to others. It is also a seemingly costly, altruisticbehavior, so caller should take into account whether their alarm calls are likely to benefit another.

3. Wild chimpanzees produce 2 basic types of vocal alarms: • Loud alarm barks & “SOS” screams – typically given after identifying a deadly ambush threat (leopard, python, neighboring chimp group) • or quiet “alert hoos” – typically givento less serious ambush predators (snares, highly poisonous vipers, fresh feces from neighboring chimps or leopards). • Direct benefits (e.g., that deter predators) of alarm calls are unlikely • Alert hoos consistently elicit cryptic behavior in receivers, such as silent, cautious approach or silent avoidance behavior. • Vipers are highly camouflaged and although they prey on smaller animals, if trodden on they can respond with a deadly bite. • Alarm indicating presence and specific location of a viper beneficial to others but potentially costly to caller (might attract attention of predators or hostile chimps).

4. Question: Will chimp who has spotted a snake be more likely to alarm call if the potential recipient is unaware of the danger? • Field experiment: Subject encounters a motionless but potentially harmful viper model in the presence of others. • Study site: Budongo Forest, Uganda • Prediction: Subjects will call less when all receivers are aware of the snake compared to when at least one receiver is still naive. • Design:

5. Snake model

6. Rhinoceros viper - Bitis nasicornis Gaboon viper - Bitis gabonica • Ambush predators ofCentral and East Africa • Camouflaged among fallen leaves • and dappled light on forest floor

7. Rhinoceros viper - Bitis nasicornis Gaboon viper - Bitis gabonica • Ambush predators ofCentral and East Africa • Camouflaged among fallen leaves • and dappled light on forest floor

8. N = 33 different chimps. • Snake models presented 22 times, each presentation lasting several hours, which resulted in n = 111 encounters. • Individuals saw a snake model 1–6 times (mean = 2.8). • 22/33 chimpanzees produced alert hoos in response to seeing a model in 46 of 111 encounters (41.4%) and produced “alarm barks” in 3 of 111 encounters (2.7%). • In contrast, only 1/21 chimpanzees (1/42 encounters) produced alert hoos without seeing the snake but after hearing another group member produce alert hoos in response to the snake. • Finally, chimpanzees that had previously seen the snake but could no longer see it alarm-called in 3/18 encounters (16.7%). Motivation to produce alert hoos may thus be stimulated to some extent by seeing the snake.

9. For each trial, any individual who was able to see the snake was termed a “subject,” whereas all other individuals present within 50 m of the snake model were termed “receivers.” • Typically, a party of chimpanzees encountered the snake model as they climbed in or out of a food tree or traveled along a path. • The first chimpanzee to see the snake called the ‘(first) detector’. In these cases, subjects and receivers had no prior exposure to the snake model on that day (31 of 111 cases). • In the remaining 80 of 111 cases, subjects saw the snake after another chimpanzee had already detected the snake. • In these cases, subjects and (often) receivers had some prior exposure to the snake model, having already seen the snake or heard alert hoos. • This typically occurred when chimpanzees followed behind a detector, climbed back down a large food tree after feeding, or traveled back along the same path later in the day.

10. From the moment of the first alert hoo, all individuals within earshot (50 m) were considered “knowledgeable” (46/111 cases). • Individuals who arrived more than 3 min later and were not part of the initial party when the alert hoo was produced were considered “ignorant” (5 of 111 cases). • Those considered to have “heard” an alert hoo retained their knowledgeable status for the remainder of the day. [Do they mean for the remainder of the 2-hour trial? One trial per day?] • Thus, receivers were still considered knowledgeable if they were present when the last alert hoo was produced hours [?] earlier (mean = 17.7 min, SD = 32.4 min), even if they themselves had not seen the snake. • From the subject's perspective, others' knowledge about the snake fell into three categories: (1) ignorant receivers (no exposure to the snake), (2) partially knowledgeable receivers (heard alert hoos but had not seen the location of snake), or (3) fully knowledgeable receivers (had seen the location of snake)

11. Movies Movie S1. Subject with Ignorant Receivers Emits Alert Hoos Adult males Kato and Nick are traveling along a trail. Kato is 20 m ahead of Nick. Kato startles upon seeing the snake, which is located at the edge of the trail, and then produces a series of alert hoos. He then turns to look back at Nick, producing louder alert hoos as he turns. Nick, the alpha male, now some 10 m behind, has stopped traveling. With the louder alert hoos, Nick sits without approaching further. Later, Nick makes a detour, traveling off the road through the forest before returning to the road again after passing the position of the snake model. Nick never actually sees the snake.

12. Movie S2. Subject with first Ignorant, then knowledgeable receivers Adult male Squibbs is traveling ahead of a mother, Kutu, with her two offspring, subadult Kana and dependent infant Kathy. Squibbs sees the snake that is placed in the middle of a small path. He startles and then emits five alert hoos over 6 s. He diverts from the path, taking another route, and leaves. Kutu then arrives 12 s after Squibbs gave his final alert hoo and left. Within our classification, she and her daughter Kana are now receivers classed in the “heard” category, having arrived within 3 min of the last alert hoo. This means that she has most likely heard the alert hoo, but because Squibbs left before she arrived, she has not seen the exact position of the half-hidden, highly camouflaged snake. Kutu diverts off the path rather than approaching the snake. She rejoins the path after 10 m and continues her journey. She does not emit any vocalizations, and neither does her daughter, Kana (10 years old), arriving 2 s after her mother leaves. Kana looks toward the snake and then follows the same trajectory as her mother.

13. Data Analysis • Subject Information (prior knowledge): Subjects may or may not have had prior knowledge of the snake (in 31 cases subject was first detector, in 80 cases subject had already heard alert hoos or seen snake). • Receiver Information (prior knowledge):Initial analysis: aware of snake or not. Later analysis: (2) had heard hoos but not seen snake, (3) had seen snake (and heard hoos). • Subject Information and Receiver Information were highly correlated (Spearman's r > 0.60), i.e., if subject was aware of snake, receivers generally were too [though not always, otherwise we wouldn’t have our ‘natural experiment!]

14. Data Analysis – Predictor Variables • Variables besides subject and receiver information that might influence subjects' calling behavior included • level of threat to subject (e.g., distance to snake) • audience composition (e.g., strong social bond partner) • level of threat to receivers (e.g., distance to snake) • prior exposure to alert hoos • prior exposure to snake • Stats: Ran two sets of generalized linear mixed models, GLMMs (one for Subject Information, one for Receiver Information). In each set of models, predictor variables tested against one of two response variables (Call or Not Call and Number of Calls).

15. Figure 1 17% 30% 70% Figure 1. Influence of Receiver Information on Subjects' Likelihood to Emit Alert Hoos upon Seeing the Snake Model. Black indicates no alarm calls produced; gray indicates at least one alarm call produced. “Receiver information” indicates receiver ignorance or knowledge from the perspective of the subject, divided into the following three categories. “Seen” indicates knowledgeable receivers: the subject had seen all receivers see the snake model. “Heard” indicates partially knowledgeable receivers: the subject had heard an alarm call when all receivers were within 50 m of the snake model but could not have seen all receivers see the snake model. “Ignorant” indicates that the subject could not have seen all receivers see the snake and had not heard an alert hoo when all current receivers were within earshot (50 m) of the alert hoo.

17. Because Receiver Information and Subject Information were confounded, they ran a third set of models to distinguish which of these were driving calling behavior. • Specifically, they asked whether subjects' calling behavior was principally influenced by their own previous exposure, or habituation, to the snake, as opposed to their awareness of receivers' knowledge about the snake. • To control for subjects' differing levels of prior snake exposure, they only included cases where both subjects and receivers were previously aware of the snake model (and where receivers were further away from the snake than subjects and receivers were approaching the snake, n = 15 subjects, 30 cases). • Since receivers are knowledgeable, not ignorant, the key variable becomes whether receivers had seen snake themselves or had only heard alarm hoos. • To measure habituation, they included two variables: whether subjects had or had not previously produced alert hoosand how long they had been exposed to the snake.

18. Prediction: if subjects’ calling • driven by habituation, they’d be less likely to call and call less, the longer they’d been aware of the snake and if they’d already called • principally influenced by receivers’ knowledge state, subjects would call less if receivers had seen the snake (and its location), than if they’d only heard an alert hoo (and not seen snake’s location)

19. Conclusions and Implications • Chimpanzee alarm-calling behavior was driven not just by own danger but by whether others were aware or not of the danger. • The likelihood of calling was not related to the subjects' or receivers' perceived risk. • The likelihood of calling was not related to the subjects‘ habituation to snake-related stimuli. • “Implication of study is that chimpanzees keep track of information available to receivers and intentionally inform those who lack certain knowledge”.

20. “These results extend the current literature in four ways”: “Challenge current opinion that it is unlikely that nonhuman animals integrate recognition that others are ignorant with provision of the missing information through communication” [reference to M. Tomasello, J. Call, B. Hare (2003) “Chimpanzees understand psychological states - the question is which ones and to what extent”.] “Study is in line with recent studies suggesting that chimpanzees have a degree of volitional control over vocal production”. Shows that chimpanzee vocal behavior is influenced by a prosocial motivation, namely intentionally informing others of danger. Contrast with mixed support to date. “Particularly relevant to debates about the role of mental state attribution in the evolution of language. Some scientists have argued that a crucial stage in this evolution occurred when individuals began producing vocalizations with the goal of informing and thereby reducing ignorance in others”. [Reference to Pinker; Seyfarth & Cheney]