Search

Our human capacity to efficiently learn from other individuals is unparalleled in any nonhuman species. Some scholars argue that our propensity to learn socially is supported by an early-emerging expectation that communicative cues will convey generic information (Csibra & Gergely, 2011). In the current 2 studies, we examine whether this expectation about generic information is unique to humans by testing a species that readily attends to human cues-dogs. Specifically, we adapted a violation of expectation paradigm previously used with human infants to examine whether communicative cues lead dogs to selectively encode generic, kind-relevant information about objects (e.g., shape). Prior work has demonstrated that human infants are more likely to notice unexpected changes in kind-relevant information in communicative contexts (i.e., when an agent points to the object; Yoon et al., 2008). In contrast, across 2 studies (N = 136), dogs were no more likely to notice kind-relevant changes in communicative contexts than noncommunicative contexts. These findings suggest that although dogs attend to human communicative cues, such cues do not shape the way that dogs encode objects. More broadly, this finding lends support to the claim that our early-emerging generic expectation crucially supports our human capacity to efficiently learn from one another. (PsycInfo Database Record (c) 2021 APA, all rights reserved).

Dogs excel at understanding human social-communicative gestures like points and can distinguish between human informants who vary in characteristics such as knowledge or familiarity. This study explores if dogs, like human children, can use human social informants’ past accuracy when deciding whom to trust. Experiment 1 tested whether dogs would behave differently in the presence of an accurate (vs. inaccurate) informant. Dogs followed an accurate informant’s point significantly above chance. Further, when presented with an inaccurate point, dogs were more likely to ignore it and choose the correct location. Experiment 2 tested whether dogs could use informant past accuracy to selectively follow the point of the previously accurate informant. In test trials when informants simultaneously pointed at different locations (only one of which contained a treat), dogs chose the accurate informant at chance levels. Experiment 3 controlled for non-social task demands (e.g. understanding of hidden baiting and occlusion events) that may have influenced Experiment 2 performance. In test trials, dogs chose to follow the accurate (vs. inaccurate) informant. This suggests that like children, dogs may be able to use informants’ past accuracy when choosing between information sources. Supplementary Information The online version contains supplementary material available at 10.1007/s10071-021-01493-5.

It has been suggested that over the course of domestication, dogs developed the propensity to "look back" or gaze at humans when they encounter a challenging task. Unfortunately, little work to date has addressed the question of why dogs look back. To explore this issue, we conducted 3 experiments in which dogs had the option of doing something other than looking back at their owner when encountering an unsolvable task. In Experiments 1 and 2, dogs could look back or attempt an alternative puzzle. In both experiments, dogs attempted the alternative puzzle prior to looking back. In Experiment 3, when dogs encountered the unsolvable task, they could look back or attempt to solve the same puzzle using an alternate approach. As in Experiments 1 and 2, dogs attempted the alternate approach prior to looking back. Although some scholars have suggested that dogs may look back because they are overly reliant on humans, our findings suggest that dogs may instead prioritize independent exploration over looking back. (PsycInfo Database Record (c) 2021 APA, all rights reserved).

Metacognition refers to the ability to monitor one's own mental states. In the current study, we investigate whether domesticated dogs (Canis familiaris) and nondomesticated dingoes (Canis dingo) demonstrate metacognition by seeking information to remedy their own ignorance. In 2 studies, we used a naturalistic information-seeking paradigm in which subjects observed a human experimenter hiding a food reward behind an apparatus. Subjects could seek information by looking through a central window-like section of the apparatus to see where the reward was hidden. In Study 1, we tested whether dogs and dingoes were willing to seek information when interacting with the apparatus, finding that both species readily sought information when it was available to them. Study 2 provided a direct test of whether dogs and dingoes would seek information to rectify their own ignorance. We found evidence that both dogs and dingoes sought out information and were more likely do so when they did not already know where the treat was hidden. These results provide additional evidence suggesting that domesticated dogs seek information in the face of ignorance, as well as the first evidence of similar behavior in a nondomesticated canid. (PsycInfo Database Record (c) 2020 APA, all rights reserved).

We introduce a new comparison species-domesticated dogs (Canis familaris)-that can shed light on the evolutionary origins of shared reality. Given that dogs share many basic building blocks of shared reality (e.g. representing others' perceptions, emotions, and behaviors) they provide an ideal species for pinpointing unique aspects of shared reality in humans. In particular, current research with dogs underscores two aspects of shared reality that may be special to humans. First, humans may be unique in our tendency to share reality involuntarily. Second, humans may be unique in the extent to which we share reality. Although both humans and dogs share reality in one-on-one interactions, only humans share reality at the more extensive group and cultural level.

Humans evaluate other agents' behavior on a variety of different dimensions, including morally, from a very early age. For example, human infants as young as 6-months old prefer prosocial over antisocial others and demonstrate negative evaluations of antisocial others in a variety of paradigms (Hamlin et al. in Nature 450(7169):557, 2007; Dev Sci 13(6):923-929, 2010; Proc Natl Acad Sci 108(50):19931-19936, 2011). While these tendencies are well documented in the human species, less is known about whether similar preference emerge in non-human animals. Here, we explore this question by testing prosocial preferences in one non-human species: the domestic dog (Canis lupus familiaris). Given the ubiquity of dog-human social interactions, it is possible that dogs display human-like social evaluation tendencies. Unfortunately, prior research examining social evaluation in dogs has produced mixed results. To assess whether differences in methodology or training differences account for these contrasting results, we tested two samples of dogs with different training histories on an identical social evaluation task. Trained agility dogs approached a prosocial actor significantly more often than an antisocial actor, while untrained pet dogs showed no preference for either actor. These differences across dogs with different training histories suggest that while dogs may demonstrate preferences for prosocial others in some contexts, their social evaluation abilities are less flexible and less robust compared to those of humans.

Human children and domesticated dogs learn from communicative cues, such as pointing, in highly similar ways. In two experiments, we investigate whether dogs are biased to defer to these cues in the same way as human children. We tested dogs on a cueing task similar to one previously conducted in human children. Dogs received conflicting information about the location of a treat from a Guesser and a Knower, who either used communicative cues (i.e., pointing; Experiments 1 and 2), non-communicative physical cues (i.e., a wooden marker; Experiment 1), or goal-directed actions (i.e., grasping; Experiment 2). Although human children tested previously struggled to override inaccurate information provided by the Guesser when she used communicative cues, in contrast to physical cues or goal-directed actions, dogs were more likely to override the Guesser's information when she used communicative cues or goal-directed actions than when she used non-communicative physical cues. Given that dogs did not show the same selective bias towards the Guesser's information in communicative contexts, these findings provide clear evidence that dogs do not demonstrate a human-like bias to defer to communicative cues. Instead, dogs may be more likely to critically evaluate information presented via communicative cues than either physical or non-communicative cues.

When learning from others, human children tend to faithfully copy – or ‘overimitate’ – the actions of a demonstrator, even when these actions are irrelevant for solving the task at hand. We investigate whether domesticated dogs (Canis familiaris) and dingoes (Canis dingo) share this tendency to overimitate in three experiments. In Experiment 1, dogs and dingoes had the opportunity to solve a puzzle after watching an ostensive demonstrator who used both a relevant action and an irrelevant action. We find clear evidence against overimitation in both species. In contrast to human children (Horner & Whiten, 2005), dogs and dingoes used the irrelevant action less often across trials, suggesting that both species were filtering out the irrelevant action as they gained experience with the puzzle (like chimpanzees; Horner & Whiten, 2005). Experiments 2 and 3 provide further evidence against overimitation, demonstrating that both species’ behavior is better characterized by individual exploration than overimitation. Given that both species, particularly dogs, show human-like social learning in other contexts, these findings provide additional evidence that overimitation may be a unique aspect of human social learning. A video abstract of this article can be viewed at: https://youtu.be/g2mRniJZ7aU.