
==== Front
Sci Rep
Sci Rep
Scientific Reports
2045-2322
Nature Publishing Group UK London

39313494
73340
10.1038/s41598-024-73340-9
Article
Domain-specific inferences about conspecifics’ skills by chimpanzees
Keupp Stefanie skeupp@dpz.eu

123
Herrmann Esther 4
1 https://ror.org/02f99v835 grid.418215.b 0000 0000 8502 7018 Cognitive Ethology Laboratory, German Primate Center – Leibniz Institute for Primate Research, Kellnerweg 4, 37077 Göttingen, Germany
2 https://ror.org/01y9bpm73 grid.7450.6 0000 0001 2364 4210 Department for Primate Cognition, Johann-Friedrich-Blumenbach Institute, Georg-August-Universität Göttingen, Kellnerweg 4, 37077 Göttingen, Germany
3 https://ror.org/02f99v835 grid.418215.b 0000 0000 8502 7018 Leibniz ScienceCampus, German Primate Center – Leibniz Institute for Primate Research, Kellnerweg 4, 37077 Göttingen, Germany
4 https://ror.org/03ykbk197 grid.4701.2 0000 0001 0728 6636 Department of Psychology, University of Portsmouth, Portsmouth, PO1 2UP UK
23 9 2024
23 9 2024
2024
14 2199626 4 2024
16 9 2024
© The Author(s) 2024
2024
https://creativecommons.org/licenses/by-nc-nd/4.0/ Open Access This article is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License, which permits any non-commercial use, sharing, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if you modified the licensed material. You do not have permission under this licence to share adapted material derived from this article or parts of it. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by-nc-nd/4.0/.
Chimpanzees collaborate with conspecifics in their daily life. However, the cognitive processes underlying partner recruitment aren’t fully understood. In the current study, chimpanzees needed to recruit a conspecific partner for either a cooperative or competitive experimental task. They spontaneously preferred to recruit cooperation partners who they have seen performing successfully before on a similar task, over partners who had failed. In contrast, the chimpanzees needed to experience the consequences of competing against co-action partners before settling on a preference for the unsuccessful partner. This divergent pattern may be due to increased cognitive demands of competitive compared to cooperative tasks. Despite the observed differences of social information use in our cooperative and competitive experimental tasks, the findings are exciting as they extend our knowledge of chimpanzee’s social evaluation abilities by showing that they can draw domain-specific inferences about conspecifics’ skills.

Keywords

Social evaluation
Chimpanzees
Inference
Learning
Two informant paradigm
Competition
Cooperation
Subject terms

Psychology
Social evolution
http://dx.doi.org/10.13039/501100001659 Deutsche Forschungsgemeinschaft 425330201 Keupp Stefanie issue-copyright-statement© Springer Nature Limited 2024
==== Body
pmcIntroduction

Chimpanzee social decision making is crucial in their daily lives, ranging from cooperative activities such as territory defense1,2, affiliative interactions such as grooming3, competitive interactions such as aggressive fights for dominance status4, to simply deciding with whom to travel or who to observe cracking nuts5. In each case, knowledge about the involved individuals is crucial for making optimal decisions. Previous research has shown that chimpanzees are excellent social decision-makers6–8. For example, they form strategic alliances and invest in relationships that benefit them for rank acquisition in the future9–12, they learn from knowledgeable and dominant individuals13–15, trust friends more than non-friends16, can use both direct and indirect information to evaluate other social agents, such as human experimenters17–19 and conspecifics13,20, and recruit the best collaboration partner when they need help during a cooperation task20. Although these social decisions appear to be rational and adaptive, we don’t know which reasoning process is driving the chimpanzees’ decisions; for example, whether they chose a collaboration partner (or learning model) because they view the other as a competent individual per se, or because they previously experienced successful cooperations with this individual (or positive outcomes when copying them) and try to re-create the conditions they associate with a successful outcome. However, the type of inference that individuals draw from past behavior of a social agent to predict the agent’s future behavior, might matter for the outcome of their choices. For example, a close affiliate might be a good choice of partner for situations that require social support, but at the same time might be a bad model choice for learning a new skill because they are not very proficient at it.

Recent work in developmental psychology has focused on theorizing about the inferential processes underlying social selectivity and its development in children21–23. Based on empirical work showing that children begin to be selective about whom they attend to, imitate, and trust, and that they use epistemic and social cues to inform their social decisions24–32, three candidate processes have been put forward21,22. First, an individual might predict another agent’s future behavior based on behavior matching. In behavior matching, an agent assumes that a model shows behavioral consistency over time. Thus, someone who has demonstrated, say, accuracy in object-labelling in the past will provide accurate labels in the future. An inference based on behavior matching is relatively narrow and restricted to the behavior at hand (e.g., object-labelling). Second, an individual might predict another agent’s future behavior based on a global impression. Inference by global impression formation lets the subject predict a wide range of behaviors based on a general impression of the other agent. In general impression formation, an individual conceives a model as positive/negative in undifferentiated ways (also referred to as “halo/pitchfork” effects), leading to wider competence predictions than warranted. For example, the individual might infer that someone who is strong is also an accurate object-labeler. Third, an individual might make rational trait-like inferences about another agent’s typical behavior: Trait-reasoning allows to predict trait-related behaviors while avoiding overgeneralization. It requires a more complex inference process than behavior matching and global impression formation. An individual who engages in trait reasoning succeeds in a two-component process33. The first component involves mapping observed behavior to a trait. The second component involves mapping the assigned trait to future behavior. Rational trait-reasoning allows to make adequate predictions in flexible and context specific ways. Younger children rely on simpler decision heuristics and, as they grow older, they gradually learn to use more sophisticated reasoning processes34–37. Children begin making trait-like inferences from around four years of age22.

The goal of the current study was to investigate if chimpanzees attribute skills to others based on inference-based reasoning or rely on outcome-based individual learning, i.e., learning by experiencing successes and failures in interactions with a partner. With the study setup, we could not unequivocally differentiate between the possibilities that chimpanzees formed a general impression of a partner or used trait-based reasoning; therefore, our interpretation will focus on the cognitively simpler inference type, i.e., narrow behavior matching. In case we find support for inference-based reasoning beyond narrow behavior matching as a cognitive underpinning of chimpanzee social selectivity, further disentangling trait-based reasoning and general impression formation will be an exciting topic for future studies.

We experimentally investigated if chimpanzees rely on repeated personal experience from direct interactions with others to make optimal partner choices or if they spontaneously pick cooperators and competitors based on whether they have seen potential partners succeed or fail at a task. Importantly, the experimental scenario was not about socially learning to perform a task by means of observing others, but rather about who the individuals choose as an interaction partner. We presented chimpanzees with two kinds of co-action situations, in which success depended on a partner’s task performance. In the cooperation setting (Experiment 1), subjects were only successful when working with a skillful conspecific partner. In the competition setting (Experiment 2), subjects were only successful when competing against an unskilled conspecific partner. In each case, subjects decided with who of two potential partners they wanted to co-act. Crucially, they had not co-acted with any of the partners in this particular setup, previous to the respective test session. Following the two-informant paradigm from developmental psychology26,27,38, subjects were first provided with information about the partners, and could then choose with which partner they wanted to co-act (note that children were often asked to verbally indicate which of two models they endorse rather than actually co-acting with them). Specifically, the chimpanzees witnessed each partner performing a puzzle box task equally often. One partner was always successful, while the other partner failed most of the time (success/failure was manipulated by the experimenter). During the test sessions following this information sampling, either the same puzzle boxes or novel puzzle boxes were presented. Including both familiar box condition and novel box condition in Experiment 1 allowed us to get a first impression of the breadth of chimpanzees’ inferences regarding partner competence.

In Experiment 1, the subject had to choose a cooperation partner. In this case, the optimal choice was to pick the partner who could previously solve the task successfully and reliably. In the “familiar task” condition, the same puzzle box was used as during the previous observation phase. In the “novel task” condition, a different puzzle task was used, which meant that subjects had never seen the partners working on the task. If chimpanzees rely on direct personal experience to make optimal partner choices, we would expect them to choose randomly, at first, and only develop a preference for the skilled partner after repeatedly experiencing successes and failures in direct interactions with the partners. This pattern should be unaffected by task type (familiar or novel). If they use inference-based reasoning they should be able to transfer their knowledge of the partners from the observation phase into the test phase. We would expect them to prefer the skilled partner immediately and throughout the test sessions. If they only show this pattern in the familiar condition, this would indicate a very narrow type of behavior matching at work. If they also show this pattern in the novel condition, this would indicate that they formed a more general impression of the partners’ competence to solve puzzle boxes, i.e., to draw domain-specific inferences about conspecifics’ skills.

In Experiment 2, subjects had to choose a competitor. The optimal strategy to maximize one’s rewards was to pick the partner who previously failed the task. The rationale is equivalent to Experiment 1: An immediate preference for the unskilled partner would indicate that chimpanzees infer the likely outcome of the race from their knowledge of how the partners previously performed the solo version of the task. In contrast, if they only develop a preference for the unskilled partner with increasing trial number, this indicates that they rely on experiencing interaction outcomes directly. In Experiment 2, we only used the familiar task.

Our results will extend existing findings in important ways by clarifying whether chimpanzees rely on experiencing interaction outcomes directly to develop a social preference, or if they use inference-based reasoning to predict others’ behavior. We had no particular predictions with regard to whether the chimpanzees would use different strategies in cooperation and competition settings. On the one hand, they have to navigate both types of situations on a daily basis and thus should be competent in succeeding in both experiments. On the other hand, each setting comes with its own cognitive and social challenges, including issues of dominance, social preferences, and perspective-taking demands. We elaborate on these aspects when introducing the methodologies of Experiments 1 and 2, and in the discussion.

Results

Experiment 1

We tested eleven chimpanzees in a familiar box condition and a novel box condition. They were randomly assigned to be tested in the order familiar-novel (n = 5) or novel-familiar (n = 6). All subjects were tested with two conspecific partner pairs, whereby we assigned one partner individual the role of skillful and the other partner individual the role of unskilled partner in each pair. Chimpanzees participated either in the role of subject or partner but not both (see also Table S4 in the supplementary information). We ran eight trials per task (familiar, novel) with each of the two partner pairs, resulting in each subject making 32 choices in total (16 per partner pair). To estimate whether the individuals preferred to choose the competent partner in each task and how their choices developed over the course of sessions and trials, we fitted a Generalized Linear Mixed Model (GLMM)39 with binomial error structure and logit link function40. To control for their potential effects, we included subjects’ initial partner preference as a further fixed effect into the model. The model comprised random intercept effects for the ID of the pair and the ID of the individual tested and all theoretically identifiable random slopes whereby factors entering the random slopes part were manually dummy coded and then centered to a mean of zero. This full model was compared to a null model lacking task and the interactions with it in the fixed effects part. The model was fitted in R (version 4.2.2) using the function glmer of the package lme441 (version 1.1-31).

Chimpanzees overall preferred the skilled partner; they chose the skilled partner 247 times and the unskilled partner 105 times (Figs. S7 and S8 show a detailed breakdown of individual choices in each trial). There was no significant effect of the predictor variables (χ2 = 1.328, df = 3, p = 0.723), indicating that neither condition, nor its interactions with presentation order and trial number affected the chimpanzees partner choice significantly (for detailed results of the model see Table 1). We then removed the interactions, which were not significant, to be able to check whether any of the main effects involved in them were significant. This revealed none of them to be significant (Table 2). Plotting the fitted full model with respect to trial number and task illustrates that the tested individuals on average preferred to choose the skilled partner, irrespective of trial number and task (Fig. 1). The absence of a clear statistical preference for the optimal partner at trial 1, as apparent from the confidence interval range of the intercept estimate (see Table 2), complicates interpretation of the results. Specifically, while visual inspection of the data indicates a spontaneous preference for the skilled partner, we cannot exclude the possibility that chimpanzees relied on that first trial to form their impression of who is the better partner to work with. To help with interpreting this pattern, we conducted an additional post hoc analysis to assess if chimpanzees were influenced by the outcome of their previous choices, i.e., whether they were maybe following a win-stay/lose-shift strategy. This analysis confirmed that partner choice was not affected by success in the previous trial (probability to switch after success = 0.304, probability to switch after failure = 0.225; χ2 = 1.990, df = 1, p = 0.117), indicating the subjects entered the test phase prepared to choose the better partner while influence from direct experience was small.Table 1 Results of the full model for experiment 1 (estimates and standard errors, together with confidence limits, significance tests, and the range of estimates obtained when dropping levels of grouping factors one at a time).

Term1	Estimate	SE	CLLower	CLUpper	χ2	df	P	min	max	
(Intercept)	0.760	0.612	− 0.438	2.142				0.215	1.022	
Task	0.418	0.520	− 0.616	1.586	0.268	0.618	
Trial	0.742	0.584	− 0.523	2.105	0.494	1.014	
Order	− 0.268	0.953	− 2.320	1.682	− 0.933	0.882	
Preference	− 0.679	0.356	− 1.522	0.085	2.937	1	0.087	− 0.914	− 0.242	
Pair	0.082	0.419	− 0.880	1.038	0.039	1	0.844	− 0.032	9.726	
Task × Trial	− 0.304	0.852	− 2.246	1.427	0.125	1	0.723	− 0.884	0.093	
Task × Order	− 0.726	0.627	− 2.116	0.653	1.203	1	0.273	− 1.121	− 0.481	
1Initial preference was z-transformed to a mean of 0 and a standard deviation (sd) of 1; mean (sd) of the original variables was 3.227 (2.068). Task was dummy coded with the novel task being the reference level, condition order was dummy coded with novel-familiar being the reference level, and Partner pair was dummy coded with 1 being the reference level.

Table 2 Results of the reduced model for experiment 1 (estimates and standard errors, together with confidence limits, and significance tests).

Term1	Estimate	SE	CLLower	CLUpper	χ2	df	P	
(Intercept)	0.960	0.579	− 0.127	2.256		
Task	0.012	0.314	− 0.667	0.661	0.001	1	0.970	
Trial	0.618	0.401	− 0.202	1.575	2.379	1	0.123	
Order	− 0.643	0.897	− 2.471	1.088	1.169	1	0.280	
Preference	− 0.680	0.355	− 1.567	0.003	3.312	1	0.069	
Pair	0.083	0.421	− 0.895	1.044	0.038	1	0.845	
1Initial preference was z-transformed to a mean of 0 and a standard deviation (sd) of 1; mean (sd) of the original variables was 3.227 (2.068). Task was dummy coded with the novel task being the reference level, condition order was dummy coded with novel-familiar being the reference level, and Partner pair was dummy coded with 1 being the reference level.

Fig. 1 Probability to choose the skilled partner in the cooperation task. Note. Depicted is the probability to choose the skilled partner as a function of trial number, separately for the two tasks (dark pink: novel, light orange: familiar). Each point represents the probability to choose the skilled partner combined for both partner pairs (i.e., 16 trials). The dashed lines and the their surrounding polygons depict the fitted model with 95% confidence limits. The model depicted is the is the full model, but with all predictors other than trial number and task being centered to a mean of zero.

Experiment 2

Ten chimpanzees participated as subjects in Experiment 2. They were tested with four conspecific partner pairs. Chimpanzees participated either in the role of subject or partner but not both. Only the slider task was used, hence we only presented the familiar condition, in Experiment 2. Subjects received eight test trials with each partner pair, resulting in a total of 32 choices per subject. Order of partner pair was pseudo-randomized across subjects such that no more than two subjects received the same order (see also Table S5 in the supplementary information).

To estimate whether the individuals preferred to choose the unskilled partner and how their choices developed over the course of sessions and trials, we fitted a GLMM with binomial error structure and logit link function. We included session number, trial number, and their interaction as fixed effects into the model. To control for their potential effects, we also included partner preference and prior experience as fixed effects into the model. To account for having repeated observations of the same individuals in the data, we included individual ID as a random intercept effect. As each trial involved a competent and an incompetent partner, each of these tested several times with a given individual and also with other individuals, we also included the ID of skilled and the unskilled partner as random intercept effects. To avoid an ’overconfident’ model and keep type I error rate at the nominal level of 0.05, we included all theoretically identifiable random slopes42,43, namely the entire of fixed effects within individual ID and those of trial number, session number, their interaction, and partner preference within the IDs of the competent and the incompetent partner. We determined the significance of individual fixed effects by dropping them from the model one at a time (R function drop1). This test utilized a likelihood ratio test44. We estimated whether the individuals’ probability to choose the unskilled partner was significantly above chance by determining 95% the confidence interval of the fitted model, separately for each combination of session number and trial number, and then checking whether it comprised chance level (i.e., the value of 0.5). The sample analyzed with this model comprised a total of 304 trials, conducted with 10 subjects, tested with 9 skilled partners and 8 unskilled partners.

Subjects overall preferred the unskilled partners; they chose the unskilled partners a total of 215 times and the skilled partners a total of 89 times (for a detailed breakdown of individual choices in each trial, see Fig. S9). Plotting the fitted full model with respect to trial and session number (Fig. 2) illustrates that the tested individuals developed a preference for the unskilled partner of the current pair over time (effect of trial in reduced model: χ2 = 2.364, df = 1, p = 0.004; see Tables 3 and 4 for model results of full and reduced model). This pattern disappeared with increasing session number. A post hoc analysis confirmed that partner choice was significantly affected by success in the previous trial (probability to switch after success = 0.222, probability to switch after failure = 0.617; χ2 = 12.706, df = 1, p < 0.001), indicating they mostly learned from direct experience.Fig. 2 Probability to choose the unskilled partner in the competition task. Note. Probability to choose the unskilled partner as a function of session number and trial number. The dashed lines and grey polygons depict the fitted model and its 95% confidence limits. The model depicted is the full model, but with all predictors other than trial number being centered to a mean of zero.

Table 3 Results of the full model for experiment 2 (estimates and standard errors, together with confidence limits, significance tests, and the range of estimates obtained when dropping levels of grouping factors one at a time).

Term1	Estimate	SE	CLLower	CLUpper	χ2	df	P	min	max	
(Intercept)	0.888	0.791	− 0.480	2.671				− 0.054	1.444	
Trial	2.246	1.09Z	0.065	5.002	0.868	3.742	
Session	1.245	1.335	− 4.265	1.028	− 2.316	0.025	
Prior experience	− 0.777	0.862	− 2.340	0.724	0.797	1	0.372	− 1.296	− 0.281	
Partner preference	0.378	0.530	− 0.679	1.601	0.496	1	0.481	0.064	0.937	
Trial × Session	0.410	2.482	− 4.504	5.842	0.033	1	0.856	− 1.427	3.353	
1Initial preference was z-transformed to a mean of 0 and a standard deviation (sd) of 1; mean (sd) of the original variables was 3.227 (2.068). Task was dummy coded with the novel task being the reference level, condition order was dummy coded with novel-familiar being the reference level, and Partner pair was dummy coded with 1 being the reference level.

Table 4 Results of the reduced model for experiment 2 (estimates and standard errors, together with confidence limits, and significance tests).

Term	Estimate	SE	CLLower	CLUpper	χ2	df	P	
(Intercept)	0.853	0.765	− 0.403	2.459		
Trial	2.364	0.808	0.806	4.778	8.111	1	0.004	
Session	− 1.110	1.086	− 3.259	0.589	1.323	1	0.250	
Prior experience	− 0.758	0.869	− 2.324	0.697	0.763	1	0.382	
Partner preference	0.380	0.528	− 0.697	1.587	0.517	1	0.472	
Indicated are the fixed effects estimates, together with their standard errors, 95% confidence limits, and significance tests; session number and trial number were scaled to a range from 0 to 1; partner preference was z-transformed to a mean of 0 and a standard deviation (sd) of 1; mean and sd of the original variable were 1.605 and 0.814, respectively; prior experience was dummy coded with ’no’ being the reference level.

Discussion

The current study aimed to find out more about whether chimpanzees rely on personal experience from direct interactions with others to choose the better of two partners for collaboration and competition, or whether they attribute skills to others based on their prior observations of potential partners succeeding or failing at a task.

In both studies, estimated trial 1 performance was not different from chance. At the same time, the chimpanzees overall preferred the respective optimal partner. In the cooperation experiment, the preference for the skilled partner was not significantly affected by trial number and choices followed very similar patterns in familiar and novel condition. The chimpanzees’ choices were not significantly affected by the outcome of the previous trial. In contrast, in the competition experiment, the preference for the unskilled partner developed over the course of the trials, demonstrating a clear learning process influenced by the direct outcomes of their interactions. Subjects switched partner significantly more often after failure than after success. Altogether, the emerging picture indicates that when recruiting a conspecific cooperation partner, chimpanzees seemed to be set up for higher initial success without considering outcome information from previous trials. They showed the same almost immediate preference for the skilled partner when they were confronted with a collaborative task involving boxes they have never seen the others interacting with before. This pattern indicates that they attributed a general (in)competence to solve puzzle boxes to the partners and that they did so based on their previous observations of the partners. When confronted with a competitive co-action task, chimpanzees learned to choose the less competent of two partners with increasing trial number by using a win-stay/lose-shift strategy. This pattern is in line with our prediction that an impression of the partners’ utility was formed by means of learning from successes and failures rather than by any of the inference types (behavior matching, general impression, or trait reasoning).

The discrepancy between the chimpanzees’ performance in the two experiments is surprising at first glance, given that they have to navigate both cooperative and competitive situations on a daily basis and thus should be similarly able to succeed in both experiments. But some considerations are of interest, in this respect. First, partner selectivity in competitive tasks is more cognitively demanding than recruiting a cooperation partner. Competitive tasks pose more complex perspective-taking demands45,46 because the subject has to consider that their own goal and the goal of the recruited partner are in conflict. Conversely, in a cooperative context, the perspective of subject and the recruited partner align. While there is plenty of evidence that chimpanzees have so-called Level 1 perspective taking skills47–53 it remains unclear to what extent they possess higher-level understanding of perspectives54,55. For example, they failed various attempts to test their false-belief understanding51,56–58, and their choices were guided by their own preference rather than Level 2 perspective taking in a competitive choice game59. Consequently, chimpanzees may have struggled with the extra cognitive step in the current competition task and needed the additional personal experience of going empty-handed to form partner-outcome associations that led them to avoid the skillful individuals.

Second, the chimpanzees had more difficulties to understand the competitive setup compared to the cooperation setup, during familiarization stages, which may indicate that carry-over effects have interfered in the second experiment. Despite having passed the comprehension check, subjects might still have struggled to overcome their previously successful strategy to choose a skillful partner. We didn’t counterbalance the presentation order of the two experiments because we first wanted to establish that chimpanzees can, in principle, use inference-based reasoning to pick optimal co-action partners, in a context where they had already shown optimal partner choice based on personal experience, i.e., in cooperative settings e.g.,20. Importantly, if the competition task in our study was affected by a carry-over effect, this effect would be more likely due to a generic strategy of having learned that picking skilled individuals is beneficial rather than relying on partner individual specific strategies. This is because only one pair of partners was re-used in the second experiment. However, we neither see subjects responding according to their preference in the first experiment in systematic ways, nor do we find any indication of particular partner individuals driving the results, as apparent by the absence of significant random intercepts effects of partner ID. To clarify if carry-over effects interfered with partner selectivity or if competitive conditions are indeed more difficult to parse for chimpanzees, future experiments should consider to present the competitive condition first or, if a sufficiently large sample size is available, counterbalance the order of cooperative and competitive tasks.

Third, the decreased performance in the competitive condition may be related to differences in test procedures. Particularly, in Experiment 2, we presented fewer observation events and accommodated all test steps (initial preference assessment, information sampling, and test trials) into a single session, for each of the four partner pairs. We suspect the chimpanzees may have lost motivation during this relatively long procedure, which required a lot of passive observing and not so much active acting. Motivation loss would also explain why they chose at random, during the last session.

Fourth, social preferences may have a greater influence on partner selectivity and behavior during the competitive test condition than during the cooperative task. That’s because it’s riskier to race against a dominant or otherwise not closely affiliated individual (and potentially even win and receive food while the other goes empty-handed) than it is to interact with them in a cooperative task. Although the individuals were separated from each other during the test procedures, they knew they would later spend the rest of the day in the forest together, where dominant individuals may retaliate for “inappropriate” behaviors during the test situation, such as a subordinate receiving food that the dominant individual wants. This explanation can certainly not explain the overall pattern of partner choice in Experiment 2, as chimpanzees’ partner selectivity towards the end of test sessions 1–3 suggests a strong preference for the incompetent partner. However, the choice pattern may indicate that initial reluctance to choose according to one’s knowledge about the partners can be overcome by repeated direct experience of how each partner choice plays out.

In summary, the chimpanzees in our study used a mix of strategies in their partner selectivity in cooperation and competition tasks. They mostly relied on learning from direct experiences with interaction partners in the competition experiment, following a win-stay/lose-shift strategy. In contrast, the picture is more ambiguous for the cooperation experiments, where we found indications that they used inference-based reasoning to inform their social decisions and generalized partner skill to a novel task. Whether these findings can be confirmed with other study groups and domains of competence and whether chimpanzees can engage in cognitively more complex trait reasoning are exciting topics for upcoming studies.

Methods

Ethics statement

We tested chimpanzees from Ngamba Island Chimpanzee Sanctuary, Uganda (https://ngambaisland.org/). These are orphaned chimpanzees who were rescued from the illegal bushmeat and pet trade. Throughout the day the entire group (54 individuals) has access to a 95-acre secondary forest on the island to forage and roam freely. The group is additionally fed four times a day with fruits, vegetables, posho (maize flour dish) and porridge; water is available ad libitum. At night the chimpanzees sleep in a large holding facility (542 m2) consisting of nine rooms with inter-connecting corridors or sliding doors. Testing took place in familiar rooms in the holding facility. The chimpanzees were never food-deprived for this study and could stop participating at any time by leaving the testing area and approaching the door to the forest. These chimpanzees frequently participate in cognitive-behavioral testing and are familiar with different experimental setups. The current research was approved by the Chimpanzee Sanctuary and Wildlife Conservation Trust (CSWCT) as well as the Uganda Wildlife Authority (UWA/COD/96/05) and the Uganda National Council for Science and Technology (NS360). The conducted research adheres to the guidelines for the ethical treatment of nonhuman animals set down by the Association for the study of Animal Behaviour (ASAB Ethical Committee/ABS Animal Care Committee, 2023) and to the EU directive 2010/63/EU and national laws. More information on the study population is provided in the supplementary information.

Procedure experiment 1

We used two types of puzzle boxes that were mounted vertically at the bars of the testing rooms (see Fig. 3). Each box required to navigate a ball (Ø 2.5 cm) from top to bottom and avoid traps and obstacles on the way. The chimpanzees could move the ball by using their fingers or a stick, and operating sliders and a cog wheel (for a setup overview, see Fig. 4).Fig. 3 The two puzzle box tasks. Note. Left: the slider task required to move the ball by operating three levels of sliders correctly and avoid the ball being trapped in case one passage was blocked. Middle: the cogwheel task required to move the ball with a stick and turn the wheel in the correct direction to avoid any inserted block. Right: The cooperation setup with open door between two testing rooms, during familiarization. Both parts of the setup must be solved successfully to obtain a reward in exchange for the ball: The experimenter inserted the ball into the upper box, chimpanzees had to first operate the upper box, the ball rolled through a transparent hose into the lower box, chimpanzees moved to the other room and operated the lower box, and finally the ball fell into an aluminum tray and could be collected by the experimenter. Slider task and cogwheel task were never installed mixed within one and the same setup, i.e., both partners had to solve the same puzzle box in a given trial.

Fig. 4 Schematic depiction of the room setup for Experiment 1 and Experiment 2. Note. Left: Cooperation setup (Exp. 1). Subject and both partners were present in adjacent rooms. At each setup location, food rewards were in place for both subject and partner (diluted juice, peanuts); one ball was required; the experimenter received the ball from the subject and inserted it into Box 1 of the chosen cooperative setup; both individuals got rewarded in case they were successful, i.e., if the subject picked the skilled partner. Right: Competition setup (Exp. 2). Subject and both partners were present. At each location, exactly one piece of food was in place (a large piece of apple). Two balls were required: one was already in place at each setup, the other had to be handed over to the experimenter by the subject; once the experimenter received the ball from the subject at the chosen location, she simultaneously inserted the balls into Box 1 and Box 2 of the chosen setup. The chimpanzee who succeeded faster received the reward.

Prior to the test, the chimpanzees went through an extensive familiarization phase where we ensured they understood all aspects of the test environment. Specifically, they learned to operate the individual boxes, to navigate the cooperative setup (two boxes connected with a transparent hose) by moving between the adjacent rooms and operating the boxes sequentially, and to exchange the token ball with the researcher and use it to indicate their choice by transporting it to the desired location. They also had to demonstrate their principal understanding of the consequences of choosing a cooperative setup with or without a co-action partner in place by picking the correct setup, i.e., they had to reliably pick the location with a partner who could access the part of the setup in the adjacent room rather than the location with no partner in place. Eleven chimpanzees passed this final comprehension check and participated as subjects in the test. Four other chimpanzees participated in the role of partner. Figure 5 gives an overview of all familiarization steps for both experiments (for more details, please refer to supplementary information).Fig. 5 Overview of familiarization steps and test phase. Note. All subjects went through all preparatory steps (white boxes). One subject who didn’t pass the comprehension check of Experiment 1 had passed all other previous steps successfully and could be tested in Experiment 2. Partners for Experiment 1 were familiarized with both boxes (Cogs and Slider task) and the double setups, i.e., the cooperation setup. Partners for Experiment 2 only needed to be able to operate the single Slider task successfully. Experimental phase is depicted in grey boxes. In Experiment 1, the three experimental steps were run with two partner pairs and both tasks per pair; in Experiment 2, the experimental steps were run with four partner pairs and only the Slider task.

During the test phase, the subject and two conspecific partners were present in three adjacent rooms, whereby the subject occupied the middle room. Two cooperation setups were mounted—one setup spanning left and middle room and the other setup spanning right and middle room. Doors between the rooms were closed, such that only half of each setup (specifically, the lower box) was accessible for the subject. The test phase consisted of an initial preference assessment, an information sampling phase, and the test sessions proper.

In a first step, we assessed subjects’ prior preference for any of the partners, during eight trials. This step required to choose between the partners by carrying the small ball towards one of the setups and to hand it over to the experimenter (a procedure with which subjects were previously familiarized with). The experimenter rewarded both individuals but, importantly, individuals didn’t get to perform the actual task. For five subjects, the roles of skilled/unskilled partner were assigned based on the subject’s prior preference. Specifically, if one of the partners was chosen six or more out of eight times during the initial preference assessment, this partner was assigned the role of unskilled partner. If no preference was observed, we aimed to balance the role for each partner individual as best as possible (for example, if a partner had already been assigned the role of unskilled partner for several other subjects, we assigned the skilled role now, to avoid frustration). For six subjects, we didn’t consider prior partner preference and instead used a pre-allocated role assignment (note that for these six individuals, in only one case did a subject have a pronounced prior preference and continued to prefer this partner in the subsequent test session; for more details, see Table S4 in the supplementary information).

In the next step, subjects could observe both partners working on the slider task (also referred to as the “familiar” task) on three consecutive days. During this information sampling phase, they observed the allocated skillful partner succeeding 24/24 (8 successes per day) times. They observed the allocated unskilled partner succeeding 3/24 times (1 success and 7 failures per day). We ensured the partners’ successes and failures by secretly manipulating the boxes accordingly, in each trial. Each subject then received two test conditions on the following day (order of condition counterbalanced between subjects; five subjects received the familiar condition first). In the familiar condition, the slider task was mounted, i.e., the task which they had observed the partners working on during three previous days. In the novel condition, the cogwheel task was mounted, i.e., a task that subjects had never seen the partners working on. Importantly, subjects themselves were familiar with both tasks and had experienced successes and failures at each of them. We ran both test conditions (eight trials per condition) on the same day, with a break of at least 30 min between them (breaks ranged roughly from 30 to 60 min, depending on test logistics). During a test trial, subjects chose a cooperation partner by transporting the ball to the setup of their choice and handing it over to the experimenter. The experimenter inserted the ball on the partner’s side of the setup and stepped back to let the chimpanzees operate the boxes without interference. If the subject chose the skilled partner, the partner’s box was set to success and, consequently, the partner successfully moved the ball through their box, the ball rolled through the transparent hose to the subject’s side, and the subject completed the task. Both individuals received a reward (peanuts and diluted juice) from the experimenter. If the subject chose the unskilled partner, the partner navigated the ball into a trap (because the box was set to failure) and it never reached the subject’s side. The experimenter removed the ball and left without providing a reward. Importantly, all individuals had ample experience with the consequences of failed attempts—during the prior weeks of familiarization, they experienced numerous times that the experimenter only rewards successful performance, namely when the ball is collected from the aluminum tray at the bottom the box.

We repeated the test phase with a second pair of partners, such that each subject completed a total of 16 trials per condition. Order of partner pairs was the same for all subjects: Partner pair 1 were two adult males and partner pair 2 were two adult females. Partners were chosen in close consultation with the senior animal caregivers. For each triplet (subject and partner pair), the caregivers agreed that it constitutes a balanced choice for the subjects between the two partners. This evaluation was based on the caregivers’ long-term experience with the chimpanzee group. Due to a limited number of chimpanzees who were similar in age, tolerant enough to function as partner in the current study, and willing to participate in behavioral studies in general, we could not balance further factors such as age and rank.

Subjects’ choices were coded live and 25% of the data was additionally coded from video by a second person who was blind to the hypothesis of the study. Coder agreement was perfect.

Procedure experiment 2

The competition experiment was conducted in the same population and we aimed to test the same individuals who had already participated in the cooperation experiment (for an overview of procedural steps and setup, see Figs. 4 and 5). For six of them, there was a break of around five months between the studies and for five of them a break of about six weeks. One individual who had not passed the comprehension check of Experiment 1 was given the chance to participate in Experiment 2 because he had mastered all other steps successfully and in addition passed the comprehension check of Experiment 2. This comprehension check ensured that subjects understood that choosing the location with a partner in the adjacent room meant they will compete for a single reward (a big piece of apple) and might lose (which we ensured to happen 100% of the time). One previous subject didn’t pass the comprehension check of Experiment 2 and another didn’t return reliably from the forest in the evenings, during the time of data collection for Experiment 2. This resulted in a sample size of ten chimpanzees, nine of who also participated in Experiment 1. Nine additional chimpanzees participated in the role of a partner. For partner role assignment, we had the following rules: First, if a subject had encountered a partner individual already during familiarization with the competition procedure, we assigned this partner the role of unskilled partner, irrespective of the subject’s choices during initial preference trials, to ensure that the subject had to re-assess the specific partner roles and can’t succeed by basing their decision on previous direct experience with the partner. Second, if a subject didn’t have previous experience with the partners, the partner who was preferred during initial preference trials was assigned the role of skilled partner; otherwise, we aimed to balance the roles between partners as best as possible to keep frustration of partner individuals at a minimum (for more details, see Table S6).

We only presented the familiar condition, using the slider task, because Experiment 1 showed that performance wasn’t affected by task type. All subjects remembered without difficulties how to operate the box as well as the need to exchange the ball with the experimenter. In contrast to the cooperation task, the two boxes were not connected with a hose. Instead, they were operated separately and rewards were given only to the individual who succeeded or who was faster in succeeding. We familiarized the chimpanzees with the new co-action setup using human partners and conspecifics (for details of familiarization, comprehension check, and partner role assignment, see supplementary information). Based on our experience in Experiment 1, we condensed the procedure and fit all test steps into a single test day per subject and partner pair. Our aim was to accommodate more partner pairs to obtain a more representative sample of partners while also keeping the chimpanzees motivated to continue to participate. In Experiment 1, all chimpanzees showed sufficient flexibility to overcome any initially existing partner preference in response to the experimental manipulation, hence we deemed it sufficient to present only four instead of eight initial preference trials. Furthermore, we reduced the number of information sampling events because we observed that subjects picked up on the difference between partner skills easily and might not require 24 observation events per partner. Previous research found that chimpanzees differentiated between human partners after only ten demonstrations e.g.,17. Consequently, a test session in Experiment 2 comprised four initial preference trials, ten information sampling events per partner (20 observations in total), and eight test trials. This resulted in four test days per subject (one day per partner pair). We had five different orders in which different subjects encountered the partner pairs. Similar to Experiment 1, the subject was placed in the middle room with a partner individual in the adjacent room on each side. Location of partners was mainly determined by convenience and varied on a daily basis. We ensured that each partner was in each room at least once. During information sampling, skilled partners succeeded in 100% of trials, unskilled partners always failed. In the test trials, an experimenter offered a large piece of apple as reward for the winner, at each setup location. In contrast to Experiment 1, where one experimenter was sufficient to follow the subjects to the chosen location, Experiment 2 required two experimenters to place the two rewards simultaneously in front of each location (see also supplementary information). In case the subject chose the location with the skilled partner, we ensured this partner always won the race and received the apple, while the subject went empty-handed. In case the subject chose the unskilled partner, we ensured the subject always won and received the apple. Position of experimenters alternated in blocks of two trials. This procedure was repeated for all four partner pairs.

Subjects’ choices were coded live and 25% of the data was additionally coded from video by a second person who was blind to the hypothesis of the study. Coder agreement was perfect.

Additional information

Additional information is available in the accompanying electronic supplementary materials and as part of Keupp and Herrmann60 here: https://osf.io/u5mbr/

Supplementary Information

Supplementary Information.

Supplementary Information

The online version contains supplementary material available at 10.1038/s41598-024-73340-9.

Acknowledgements

We are very thankful to J. Rukundo, T. Mukungu, and the trustees and all the staff of the Chimpanzee Sanctuary and Wildlife Conservation Trust (CSWCT) for their continuous help and support. In particular, we appreciate the hard work of the animal caregivers and their support helping us to conduct the experiments. We thank I. Ayebazibwe for help with reliability coding, R. Mundry for statistical help, and H. Brown for assisting during the experiments. We also appreciate permission from the Ugandan National Council for Science and Technology and the Uganda Wildlife Authority for allowing us to conduct our research in Uganda (permit numbers UWA/COD/96/05 and NS360). SK obtained funding from the Deutsche Forschungsgemeinschaft to conduct this project (project number 425330201).

Author contributions

S.K.: conceptualization, methodology, resources, funding, investigation, project administration, writing—original draft, writing—review and editing. E.H.: Methodology, writing—review and editing

Data availability

Data and analysis code are available here: https://osf.io/u5mbr/.

Competing interests

The authors declare no competing interests.

Publisher’s note

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
==== Refs
References

1. Goodall J The Chimpanzees of Gombe: Patterns of Behavior 1986 Belknap Press of Harvard University Press
Goodall, J. The Chimpanzees of Gombe: Patterns of Behavior (Belknap Press of Harvard University Press, 1986).
2. Watts D Mitani J Boundary patrols and intergroup encounters in wild chimpanzees Behaviour 2001 138 299 327 10.1163/15685390152032488
Watts, D. & Mitani, J. Boundary patrols and intergroup encounters in wild chimpanzees. Behaviour 138, 299–327 (2001).
3. Mielke A Flexible decision-making in grooming partner choice in sooty mangabeys and chimpanzees R. Soc. Open Sci. 2018 5 172143 10.1098/rsos.172143 30109053
Mielke, A. et al. Flexible decision-making in grooming partner choice in sooty mangabeys and chimpanzees. R. Soc. Open Sci. 5, 172143 (2018).30109053
4. Gilby IC Fitness benefits of coalitionary aggression in male chimpanzees Behav. Ecol. Sociobiol. 2013 67 373 381 10.1007/s00265-012-1457-6 23459197
Gilby, I. C. et al. Fitness benefits of coalitionary aggression in male chimpanzees. Behav. Ecol. Sociobiol. 67, 373–381 (2013).23459197
5. Matsuzawa, T., Humle, T., & Sugiyama, Y. (Eds.) The Chimpanzees of Bossou and Nimba, Vol 395 10.1007/978-4-431-53921-6 (Springer Japan, Tokyo, 2011).
6. Mitani JC Kappeler PM Van Schaik CP Reciprocal exchange in chimpanzees and other primates Cooperation in Primates and Humans 2006 Berlin, Heidelberg Springer Berlin Heidelberg 107 119
Mitani, J. C. Reciprocal exchange in chimpanzees and other primates. In Cooperation in Primates and Humans (eds Kappeler, P. M. & Van Schaik, C. P.) 107–119 (Springer Berlin Heidelberg, Berlin, Heidelberg, 2006). 10.1007/3-540-28277-7_6.
7. Slocombe KE Zuberbühler K Chimpanzees modify recruitment screams as a function of audience composition Proc. Natl. Acad. Sci. 2007 104 17228 17233 10.1073/pnas.0706741104 17942683
Slocombe, K. E. & Zuberbühler, K. Chimpanzees modify recruitment screams as a function of audience composition. Proc. Natl. Acad. Sci. 104, 17228–17233 (2007).17942683
8. Wittig RM Crockford C Langergraber KE Zuberbühler K Triadic social interactions operate across time: A field experiment with wild chimpanzees Proc. R. Soc. B Biol. Sci. 2014 281 20133155 10.1098/rspb.2013.3155
Wittig, R. M., Crockford, C., Langergraber, K. E. & Zuberbühler, K. Triadic social interactions operate across time: A field experiment with wild chimpanzees. Proc. R. Soc. B Biol. Sci. 281, 20133155 (2014).
9. De Waal F Chimpanzee Politics 2007 Johns Hopkins University Press
De Waal, F. Chimpanzee Politics (Johns Hopkins University Press, 2007). 10.56021/9780801886560.
10. Duffy KG Wrangham RW Silk JB Male chimpanzees exchange political support for mating opportunities Curr. Biol. 2007 17 R586 R587 10.1016/j.cub.2007.06.001 17686425
Duffy, K. G., Wrangham, R. W. & Silk, J. B. Male chimpanzees exchange political support for mating opportunities. Curr. Biol. 17, R586–R587 (2007).17686425
11. Enigk DK Thompson ME Machanda ZP Wrangham RW Muller MN Competitive ability determines coalition participation and partner selection during maturation in wild male chimpanzees (Pan troglodytes schweinfurthii) Behav. Ecol. Sociobiol. 2020 74 89 10.1007/s00265-020-02872-7 33776193
Enigk, D. K., Thompson, M. E., Machanda, Z. P., Wrangham, R. W. & Muller, M. N. Competitive ability determines coalition participation and partner selection during maturation in wild male chimpanzees (Pan troglodytes schweinfurthii). Behav. Ecol. Sociobiol. 74, 89 (2020).33776193
12. Watts D Conflict resolution in chimpanzees and the valuable-relationships hypothesis Int. J. Primatol. 2006 27 1337 1364 10.1007/s10764-006-9081-9
Watts, D. Conflict resolution in chimpanzees and the valuable-relationships hypothesis. Int. J. Primatol. 27, 1337–1364 (2006).
13. Horner V Proctor D Bonnie KE Whiten A De Waal FBM Prestige affects cultural learning in Chimpanzees PLoS One 2010 5 e10625 10.1371/journal.pone.0010625 20502702
Horner, V., Proctor, D., Bonnie, K. E., Whiten, A. & De Waal, F. B. M. Prestige affects cultural learning in Chimpanzees. PLoS One 5, e10625 (2010).20502702
14. Kendal R Chimpanzees copy dominant and knowledgeable individuals: Implications for cultural diversity Evol. Hum. Behav. 2015 36 65 72 10.1016/j.evolhumbehav.2014.09.002 27053916
Kendal, R. et al. Chimpanzees copy dominant and knowledgeable individuals: Implications for cultural diversity. Evol. Hum. Behav. 36, 65–72 (2015).27053916
15. Kendal RL Social learning strategies: Bridge-building between fields Trends Cogn. Sci. 2018 22 651 665 10.1016/j.tics.2018.04.003 29759889
Kendal, R. L. et al. Social learning strategies: Bridge-building between fields. Trends Cogn. Sci. 22, 651–665 (2018).29759889
16. Engelmann JM Herrmann E Chimpanzees trust their friends Curr. Biol. 2016 26 252 256 10.1016/j.cub.2015.11.037 26776735
Engelmann, J. M. & Herrmann, E. Chimpanzees trust their friends. Curr. Biol. 26, 252–256 (2016).26776735
17. Herrmann E Keupp S Hare B Vaish A Tomasello M Direct and indirect reputation formation in nonhuman great apes (Pan paniscus, Pan troglodytes, Gorilla gorilla, Pongo pygmaeus) and human children (Homo sapiens) J. Comp. Psychol. 2013 127 63 75 10.1037/a0028929 22746158
Herrmann, E., Keupp, S., Hare, B., Vaish, A. & Tomasello, M. Direct and indirect reputation formation in nonhuman great apes (Pan paniscus, Pan troglodytes, Gorilla gorilla, Pongo pygmaeus) and human children (Homo sapiens). J. Comp. Psychol. 127, 63–75 (2013).22746158
18. Russell YI Call J Dunbar RIM Image scoring in great apes Behav. Process. 2008 78 108 111 10.1016/j.beproc.2007.10.009
Russell, Y. I., Call, J. & Dunbar, R. I. M. Image scoring in great apes. Behav. Process. 78, 108–111 (2008).
19. Subiaul F Vonk J Okamoto-Barth S Barth J Do chimpanzees learn reputation by observation? Evidence from direct and indirect experience with generous and selfish strangers Anim. Cogn. 2008 11 611 623 10.1007/s10071-008-0151-6 18357476
Subiaul, F., Vonk, J., Okamoto-Barth, S. & Barth, J. Do chimpanzees learn reputation by observation? Evidence from direct and indirect experience with generous and selfish strangers. Anim. Cogn. 11, 611–623 (2008).18357476
20. Melis AP Hare B Tomasello M Chimpanzees recruit the best collaborators Science 2006 311 1297 1300 10.1126/science.1123007 16513985
Melis, A. P., Hare, B. & Tomasello, M. Chimpanzees recruit the best collaborators. Science 311, 1297–1300 (2006).16513985
21. Fusaro M Corriveau KH Harris PL The good, the strong, and the accurate: Preschoolers’ evaluations of informant attributes J. Exp. Child Psychol. 2011 110 561 574 10.1016/j.jecp.2011.06.008 21802693
Fusaro, M., Corriveau, K. H. & Harris, P. L. The good, the strong, and the accurate: Preschoolers’ evaluations of informant attributes. J. Exp. Child Psychol. 110, 561–574 (2011).21802693
22. Hermes J Behne T Rakoczy H The role of trait reasoning in young children’s selective trust Dev. Psychol. 2015 51 1574 1587 10.1037/dev0000042 26389602
Hermes, J., Behne, T. & Rakoczy, H. The role of trait reasoning in young children’s selective trust. Dev. Psychol. 51, 1574–1587 (2015).26389602
23. Sobel DM Kushnir T Knowledge matters: How children evaluate the reliability of testimony as a process of rational inference Psychol. Rev. 2013 120 779 797 10.1037/a0034191 24015954
Sobel, D. M. & Kushnir, T. Knowledge matters: How children evaluate the reliability of testimony as a process of rational inference. Psychol. Rev. 120, 779–797 (2013).24015954
24. Birch SAJ Vauthier SA Bloom P Three- and four-year-olds spontaneously use others’ past performance to guide their learning Cognition 2008 107 1018 1034 10.1016/j.cognition.2007.12.008 18295193
Birch, S. A. J., Vauthier, S. A. & Bloom, P. Three- and four-year-olds spontaneously use others’ past performance to guide their learning. Cognition 107, 1018–1034 (2008).18295193
25. Jaswal VK Neely LA Adults don’t always know best: Preschoolers use past reliability over age when learning new words Psychol. Sci. 2006 17 757 758 10.1111/j.1467-9280.2006.01778.x 16984291
Jaswal, V. K. & Neely, L. A. Adults don’t always know best: Preschoolers use past reliability over age when learning new words. Psychol. Sci. 17, 757–758 (2006).16984291
26. Koenig MA Clément F Harris PL Trust in testimony: Children’s use of true and false statements Psychol. Sci. 2004 15 694 698 10.1111/j.0956-7976.2004.00742.x 15447641
Koenig, M. A., Clément, F. & Harris, P. L. Trust in testimony: Children’s use of true and false statements. Psychol. Sci. 15, 694–698 (2004).15447641
27. Koenig MA Harris PL Preschoolers mistrust ignorant and inaccurate speakers Child Dev. 2005 76 1261 1277 10.1111/j.1467-8624.2005.00849.x 16274439
Koenig, M. A. & Harris, P. L. Preschoolers mistrust ignorant and inaccurate speakers. Child Dev. 76, 1261–1277 (2005).16274439
28. Pasquini ES Corriveau KH Koenig M Harris PL Preschoolers monitor the relative accuracy of informants Dev. Psychol. 2007 43 1216 1226 10.1037/0012-1649.43.5.1216 17723046
Pasquini, E. S., Corriveau, K. H., Koenig, M. & Harris, P. L. Preschoolers monitor the relative accuracy of informants. Dev. Psychol. 43, 1216–1226 (2007).17723046
29. Bascandziev I Harris PL In beauty we trust: Children prefer information from more attractive informants Br. J. Dev. Psychol. 2014 32 94 99 10.1111/bjdp.12022 24164592
Bascandziev, I. & Harris, P. L. In beauty we trust: Children prefer information from more attractive informants. Br. J. Dev. Psychol. 32, 94–99 (2014).24164592
30. Bernard S The boss is always right: Preschoolers endorse the testimony of a dominant over that of a subordinate J. Exp. Child Psychol. 2016 152 307 317 10.1016/j.jecp.2016.08.007 27658803
Bernard, S. et al. The boss is always right: Preschoolers endorse the testimony of a dominant over that of a subordinate. J. Exp. Child Psychol. 152, 307–317 (2016).27658803
31. Corriveau K Harris PL Choosing your informant: Weighing familiarity and recent accuracy Dev. Sci. 2009 12 426 437 10.1111/j.1467-7687.2008.00792.x 19371367
Corriveau, K. & Harris, P. L. Choosing your informant: Weighing familiarity and recent accuracy. Dev. Sci. 12, 426–437 (2009).19371367
32. Jaswal VK Kondrad RL Why children are not always epistemically vigilant: Cognitive limits and social considerations Child Dev. Perspect. 2016 10 240 244 10.1111/cdep.12187
Jaswal, V. K. & Kondrad, R. L. Why children are not always epistemically vigilant: Cognitive limits and social considerations. Child Dev. Perspect. 10, 240–244 (2016).
33. Liu D Gelman SA Wellman HM Components of young children’s trait understanding: Behavior-to-trait inferences and trait-to-behavior predictions Child Dev. 2007 78 1543 1558 10.1111/j.1467-8624.2007.01082.x 17883447
Liu, D., Gelman, S. A. & Wellman, H. M. Components of young children’s trait understanding: Behavior-to-trait inferences and trait-to-behavior predictions. Child Dev. 78, 1543–1558 (2007).17883447
34. Hermes J Behne T Rakoczy H The development of selective trust: Prospects for a dual-process account Child Dev. Perspect. 2018 12 134 138 10.1111/cdep.12274
Hermes, J., Behne, T. & Rakoczy, H. The development of selective trust: Prospects for a dual-process account. Child Dev. Perspect. 12, 134–138 (2018).
35. Heyman GD Sritanyaratana L Vanderbilt KE Young children’s trust in overtly misleading advice Cogn. Sci. 2013 37 646 667 10.1111/cogs.12020 23294130
Heyman, G. D., Sritanyaratana, L. & Vanderbilt, K. E. Young children’s trust in overtly misleading advice. Cogn. Sci. 37, 646–667 (2013).23294130
36. Jaswal VK Can’t stop believing: Inhibitory control and resistance to misleading testimony Dev. Sci. 2014 17 965 976 10.1111/desc.12187 24806881
Jaswal, V. K. et al. Can’t stop believing: Inhibitory control and resistance to misleading testimony. Dev. Sci. 17, 965–976 (2014).24806881
37. Vanderbilt KE Liu D Heyman GD The development of distrust Child Dev. 2011 82 1372 1380 10.1111/j.1467-8624.2011.01629.x 21824130
Vanderbilt, K. E., Liu, D. & Heyman, G. D. The development of distrust. Child Dev. 82, 1372–1380 (2011).21824130
38. Hermes J Behne T Bich AE Thielert C Rakoczy H Children’s selective trust decisions: Rational competence and limiting performance factors Dev. Sci. 2018 21 e12527 10.1111/desc.12527
Hermes, J., Behne, T., Bich, A. E., Thielert, C. & Rakoczy, H. Children’s selective trust decisions: Rational competence and limiting performance factors. Dev. Sci. 21, e12527 (2018).
39. Baayen RH Analyzing Linguistic Data: A Practical Introduction to Statistics Using R 2008 Cambridge University Press
Baayen, R. H. Analyzing Linguistic Data: A Practical Introduction to Statistics Using R (Cambridge University Press, 2008).
40. McCullagh P Nelder JA Generalized Linear Models 1989 Springer US
McCullagh, P. & Nelder, J. A. Generalized Linear Models (Springer US, 1989). 10.1007/978-1-4899-3242-6.
41. Bates D Mächler M Bolker B Walker S Fitting linear mixed-effects models using lme4 J. Stat. Softw. 2015 67 1 48 10.18637/jss.v067.i01
Bates, D., Mächler, M., Bolker, B. & Walker, S. Fitting linear mixed-effects models using lme4. J. Stat. Softw. 67, 1–48 (2015).
42. Barr DJ Levy R Scheepers C Tily HJ Random effects structure for confirmatory hypothesis testing: Keep it maximal J. Mem. Lang. 2013 68 255 278 10.1016/j.jml.2012.11.001
Barr, D. J., Levy, R., Scheepers, C. & Tily, H. J. Random effects structure for confirmatory hypothesis testing: Keep it maximal. J. Mem. Lang. 68, 255–278 (2013).
43. Schielzeth H Forstmeier W Conclusions beyond support: Overconfident estimates in mixed models Behav. Ecol. 2009 20 416 420 10.1093/beheco/arn145 19461866
Schielzeth, H. & Forstmeier, W. Conclusions beyond support: Overconfident estimates in mixed models. Behav. Ecol. 20, 416–420 (2009).19461866
44. Dobson AJ An Introduction to Generalized Linear Models 2002 Chapman & Hall/CRC
Dobson, A. J. An Introduction to Generalized Linear Models (Chapman & Hall/CRC, 2002).
45. Perner J Zauner P Sprung M Astington JW Baird JA What does “that” have to do with point of view? Conflicting desires and “want” in German Why Language Matters for Theory of Mind 2005 Oxford University Press 220 244
Perner, J., Zauner, P. & Sprung, M. What does “that” have to do with point of view? Conflicting desires and “want” in German. In Why Language Matters for Theory of Mind (eds Astington, J. W. & Baird, J. A.) 220–244 (Oxford University Press, 2005). 10.1093/acprof:oso/9780195159912.003.0011.
46. Priewasser B Roessler J Perner J Competition as rational action: Why young children cannot appreciate competitive games J. Exp. Child Psychol. 2013 116 545 559 10.1016/j.jecp.2012.10.008 23182381
Priewasser, B., Roessler, J. & Perner, J. Competition as rational action: Why young children cannot appreciate competitive games. J. Exp. Child Psychol. 116, 545–559 (2013).23182381
47. Bräuer J Call J Tomasello M All great ape species follow gaze to distant locations and around barriers J. Comp. Psychol. 2005 119 145 154 10.1037/0735-7036.119.2.145 15982158
Bräuer, J., Call, J. & Tomasello, M. All great ape species follow gaze to distant locations and around barriers. J. Comp. Psychol. 119, 145–154 (2005).15982158
48. Bräuer J Call J Tomasello M Chimpanzees really know what others can see in a competitive situation Anim. Cogn. 2007 10 439 448 10.1007/s10071-007-0088-1 17426993
Bräuer, J., Call, J. & Tomasello, M. Chimpanzees really know what others can see in a competitive situation. Anim. Cogn. 10, 439–448 (2007).17426993
49. Hare B Call J Agnetta B Tomasello M Chimpanzees know what conspecifics do and do not see Anim. Behav. 2000 59 771 785 10.1006/anbe.1999.1377 10792932
Hare, B., Call, J., Agnetta, B. & Tomasello, M. Chimpanzees know what conspecifics do and do not see. Anim. Behav. 59, 771–785 (2000).10792932
50. Hare B Call J Tomasello M Do chimpanzees know what conspecifics know? Anim. Behav. 2001 61 139 151 10.1006/anbe.2000.1518 11170704
Hare, B., Call, J. & Tomasello, M. Do chimpanzees know what conspecifics know?. Anim. Behav. 61, 139–151 (2001).11170704
51. Kaminski J Call J Tomasello M Chimpanzees know what others know, but not what they believe Cognition 2008 109 224 234 10.1016/j.cognition.2008.08.010 18849023
Kaminski, J., Call, J. & Tomasello, M. Chimpanzees know what others know, but not what they believe. Cognition 109, 224–234 (2008).18849023
52. MacLean EL Hare B Bonobos and chimpanzees infer the target of another’s attention Anim. Behav. 2012 83 345 353 10.1016/j.anbehav.2011.10.026
MacLean, E. L. & Hare, B. Bonobos and chimpanzees infer the target of another’s attention. Anim. Behav. 83, 345–353 (2012).
53. Tomasello M Call J Hare B Five primate species follow the visual gaze of conspecifics Anim. Behav. 1998 55 1063 1069 10.1006/anbe.1997.0636 9632490
Tomasello, M., Call, J. & Hare, B. Five primate species follow the visual gaze of conspecifics. Anim. Behav. 55, 1063–1069 (1998).9632490
54. Flavell JH Shipstead SG Croft K Young children’s knowledge about visual perception: Hiding objects from others Child Dev. 1978 49 1208 10.2307/1128761 738154
Flavell, J. H., Shipstead, S. G. & Croft, K. Young children’s knowledge about visual perception: Hiding objects from others. Child Dev. 49, 1208 (1978).738154
55. Flavell JH Everett BA Croft K Flavell ER Young children’s knowledge about visual perception: Further evidence for the Level 1–Level 2 distinction Dev. Psychol. 1981 17 99 103 10.1037/0012-1649.17.1.99
Flavell, J. H., Everett, B. A., Croft, K. & Flavell, E. R. Young children’s knowledge about visual perception: Further evidence for the Level 1–Level 2 distinction. Dev. Psychol. 17, 99–103 (1981).
56. Call J Tomasello M A nonverbal false belief task: The performance of children and great apes Child Dev. 1999 70 381 395 10.1111/1467-8624.00028 10218261
Call, J. & Tomasello, M. A nonverbal false belief task: The performance of children and great apes. Child Dev. 70, 381–395 (1999).10218261
57. Krachun C Carpenter M Call J Tomasello M A competitive nonverbal false belief task for children and apes Dev. Sci. 2009 12 521 535 10.1111/j.1467-7687.2008.00793.x 19635080
Krachun, C., Carpenter, M., Call, J. & Tomasello, M. A competitive nonverbal false belief task for children and apes. Dev. Sci. 12, 521–535 (2009).19635080
58. Krachun C Carpenter CM Call J Tomasello M A new change-of-contents false belief test: Children and chimpanzees compared Int. J. Comp. Psychol. 2010 10.46867/ijcp.2010.23.02.03
Krachun, C., Carpenter, C. M., Call, J. & Tomasello, M. A new change-of-contents false belief test: Children and chimpanzees compared. Int. J. Comp. Psychol.10.46867/ijcp.2010.23.02.03 (2010).
59. Karg K Schmelz M Call J Tomasello M Differing views: Can chimpanzees do level 2 perspective-taking? Anim. Cogn. 2016 19 555 564 10.1007/s10071-016-0956-7 26852383
Karg, K., Schmelz, M., Call, J. & Tomasello, M. Differing views: Can chimpanzees do level 2 perspective-taking?. Anim. Cogn. 19, 555–564 (2016).26852383
60. Keupp, S. & Herrmann, E. Chimpanzees evaluate conspecifics’ skills in a puzzle box task. (2024). 10.31234/osf.io/udmqj.
