Chimpanzee Faces Research Articles

An open‐source general purpose machine learning framework for individual animal re‐identification using few‐shot learning

Abstract Animal re‐identification remains a challenging problem due to the cost of tagging systems and the difficulty of permanently attaching a physical marker to some animals, such as sea stars. Due to these challenges, photo identification is a good fit to solve this problem whether evaluated by humans or through machine learning. Accurate machine learning methods are an improvement over manual identification as they are capable of evaluating a large number of images automatically and recent advances have reduced the need for large training datasets. This study aimed to create an accurate, robust, general purpose machine learning framework for individual animal re‐identification using images both from publicly available data as well as two groups of sea stars of different species under human care. Open‐source code was provided to accelerate work in this space. Images of two species of sea star (Asterias rubens and Anthenea australiae) were taken using a consumer‐grade smartphone camera and used as original datasets to train a machine learning model to re‐identify an individual animal using few examples. The model's performance was evaluated on these original sea star datasets which contained between 39–54 individuals and 983–1204 images, as well as using six publicly available re‐identification datasets for tigers, beef cattle noses, chimpanzee faces, zebras, giraffes and ringed seals ranging between 45–2056 individuals and 829–6770 images. Using time aware‐splits, which are a data splitting technique ensuring that the model only sees an individual's images from a previous collection event during training to avoid information leaking, the model achieved high (>99%) individual re‐identification mean average precision for the top prediction (mAP@1) for the two species of sea stars. The re‐identification mAP@1 for the mammalian datasets was more variable, ranging from 83% to >99%. However, this model outperformed published state‐of‐the‐art re‐identification results for the publicly available datasets. The reported approach for animal re‐identification is generalizable, with the same machine learning framework achieving good performance in two distinct species of sea stars with different physical attributes, as well as seven different mammalian species. This demonstrates that this methodology can be applied to nearly any species where individual re‐identification is required. This study presents a precise, practical, non‐invasive approach to animal re‐identification using only basic image collection methods.

Supplemental Material for Search Asymmetries for Threatening Faces in Chimpanzees (Pan troglodytes)

Supplemental Material for Search Asymmetries for Threatening Faces in Chimpanzees (Pan troglodytes)

Research and conservation in the greater Gombe ecosystem: challenges and opportunities

The study of chimpanzees in Gombe National Park, Tanzania, started by Jane Goodall in 1960, provided pioneering accounts of chimpanzee behavior and ecology. With funding from multiple sources, including the Jane Goodall Institute (JGI) and grants from private foundations and federal programs, the project has continued for sixty years, providing a wealth of information about our evolutionary cousins. These chimpanzees face two main challenges to their survival: infectious disease — including simian immunodeficiency virus (SIVcpz), which can cause Acquired Immune Deficiency Syndrome (AIDS) in chimpanzees — and the deforestation of land outside the park. A health monitoring program has increased understanding of the pathogens affecting chimpanzees and has promoted measures to characterize and reduce disease risk. Deforestation reduces connections between Gombe and other chimpanzee populations, which can cause loss of genetic diversity. To promote habitat restoration, JGI facilitated participatory village land use planning, in which communities voluntarily allocated land to a network of Village Land Forest Reserves. Expected benefits to people include stabilizing watersheds, improving water supplies, and ensuring a supply of forest resources. Surveys and genetic analyses confirm that chimpanzees persist on village lands and remain connected to the Gombe population. Many challenges remain, but the regeneration of natural forest on previously degraded lands provides hope that conservation solutions can be found that benefit both people and wildlife. Conservation work in the Greater Gombe Ecosystem has helped promote broader efforts to plan and work for conservation elsewhere in Tanzania and across Africa.

Automatic Identification of Individual Primates with Deep Learning Techniques.

SummaryThe difficulty of obtaining reliable individual identification of animals has limited researcher's ability to obtain quantitative data to address important ecological, behavioral, and conservation questions. Traditional marking methods placed animals at undue risk. Machine learning approaches for identifying species through analysis of animal images has been proved to be successful. But for many questions, there needs a tool to identify not only species but also individuals. Here, we introduce a system developed specifically for automated face detection and individual identification with deep learning methods using both videos and still-framed images that can be reliably used for multiple species. The system was trained and tested with a dataset containing 102,399 images of 1,040 individuals across 41 primate species whose individual identity was known and 6,562 images of 91 individuals across four carnivore species. For primates, the system correctly identified individuals 94.1% of the time and could process 31 facial images per second.

Reduced perceptual narrowing in synesthesia

Synesthesia is a neurologic trait in which specific inducers, such as sounds, automatically elicit additional idiosyncratic percepts, such as color (thus "colored hearing"). One explanation for this trait-and the one tested here-is that synesthesia results from unusually weak pruning of cortical synaptic hyperconnectivity during early perceptual development. We tested the prediction from this hypothesis that synesthetes would be superior at making discriminations from nonnative categories that are normally weakened by experience-dependent pruning during a critical period early in development-namely, discrimination among nonnative phonemes (Hindi retroflex /d̪a/ and dental /ɖa/), among chimpanzee faces, and among inverted human faces. Like the superiority of 6-mo-old infants over older infants, the synesthetic groups were significantly better than control groups at making all the nonnative discriminations across five samples and three testing sites. The consistent superiority of the synesthetic groups in making discriminations that are normally eliminated during infancy suggests that residual cortical connectivity in synesthesia supports changes in perception that extend beyond the specific synesthetic percepts, consistent with the incomplete pruning hypothesis.

Chimpanzee face recognition from videos in the wild using deep learning.

Video recording is now ubiquitous in the study of animal behavior, but its analysis on a large scale is prohibited by the time and resources needed to manually process large volumes of data. We present a deep convolutional neural network (CNN) approach that provides a fully automated pipeline for face detection, tracking, and recognition of wild chimpanzees from long-term video records. In a 14-year dataset yielding 10 million face images from 23 individuals over 50 hours of footage, we obtained an overall accuracy of 92.5% for identity recognition and 96.2% for sex recognition. Using the identified faces, we generated co-occurrence matrices to trace changes in the social network structure of an aging population. The tools we developed enable easy processing and annotation of video datasets, including those from other species. Such automated analysis unveils the future potential of large-scale longitudinal video archives to address fundamental questions in behavior and conservation.

Viewing heterospecific facial expressions: an eye-tracking study of human and monkey viewers

Common facial expressions of emotion have distinctive patterns of facial muscle movements that are culturally similar among humans, and perceiving these expressions is associated with stereotypical gaze allocation at local facial regions that are characteristic for each expression, such as eyes in angry faces. It is, however, unclear to what extent this ‘universality’ view can be extended to process heterospecific facial expressions, and how ‘social learning’ process contributes to heterospecific expression perception. In this eye-tracking study, we examined face-viewing gaze allocation of human (including dog owners and non-dog owners) and monkey observers while exploring expressive human, chimpanzee, monkey and dog faces (positive, neutral and negative expressions in human and dog faces; neutral and negative expressions in chimpanzee and monkey faces). Human observers showed species- and experience-dependent expression categorization accuracy. Furthermore, both human and monkey observers demonstrated different face-viewing gaze distributions which were also species dependent. Specifically, humans predominately attended at human eyes but animal mouth when judging facial expressions. Monkeys’ gaze distributions in exploring human and monkey faces were qualitatively different from exploring chimpanzee and dog faces. Interestingly, the gaze behaviour of both human and monkey observers were further affected by their prior experience of the viewed species. It seems that facial expression processing is species dependent, and social learning may play a significant role in discriminating even rudimentary types of heterospecific expressions.

Exploring attentional bias towards threatening faces in chimpanzees using the dot probe task.

Primates have evolved to rapidly detect and respond to danger in their environment. However, the mechanisms involved in attending to threatening stimuli are not fully understood. The dot-probe task is one of the most widely used experimental paradigms to investigate these mechanisms in humans. However, to date, few studies have been conducted in non-human primates. The aim of this study was to investigate whether the dot-probe task can measure attentional biases towards threatening faces in chimpanzees. Eight adult chimpanzees participated in a series of touch screen dot-probe tasks. We predicted faster response times towards chimpanzee threatening faces relative to neutral faces and faster response times towards faces of high threat intensity (scream) than low threat intensity (bared teeth). Contrary to prediction, response times for chimpanzee threatening faces relative to neutral faces did not differ. In addition, we found no difference in response times for faces of high and low threat intensity. In conclusion, we found no evidence that the touch screen dot-probe task can measure attentional biases specifically towards threatening faces in our chimpanzees. Methodological limitations of using the task to measure emotional attention in human and non-human primates, including stimulus threat intensity, emotional state, stimulus presentation duration and manual responding are discussed.

Visual discrimination of primate species based on faces in chimpanzees.

Many primate studies have investigated discrimination of individual faces within the same species. However, few studies have looked at discrimination between primate species faces at the categorical level. This study systematically examined the factors important for visual discrimination between primate species faces in chimpanzees, including: colour, orientation, familiarity, and perceptual similarity. Five adult female chimpanzees were tested on their ability to discriminate identical and categorical (non-identical) images of different primate species faces in a series of touchscreen matching-to-sample experiments. Discrimination performance for chimpanzee, gorilla, and orangutan faces was better in colour than in greyscale. An inversion effect was also found, with higher accuracy for upright than inverted faces. Discrimination performance for unfamiliar (baboon and capuchin monkey) and highly familiar (chimpanzee and human) but perceptually different species was equally high. After excluding effects of colour and familiarity, difficulty in discriminating between different species faces can be best explained by their perceptual similarity to each other. Categorical discrimination performance for unfamiliar, perceptually similar faces (gorilla and orangutan) was significantly worse than unfamiliar, perceptually different faces (baboon and capuchin monkey). Moreover, multidimensional scaling analysis of the image similarity data based on local feature matching revealed greater similarity between chimpanzee, gorilla and orangutan faces than between human, baboon and capuchin monkey faces. We conclude our chimpanzees appear to perceive similarity in primate faces in a similar way to humans. Information about perceptual similarity is likely prioritized over the potential influence of previous experience or a conceptual representation of species for categorical discrimination between species faces.

Automated face detection for occurrence and occupancy estimation in chimpanzees.

Using semi-automated ape face detection technology for processing camera trap footage requires only 2-4% of the time compared to manual analysis and allows to estimate site use by chimpanzees relatively reliably.

Robust representations of individual faces in chimpanzees (Pan troglodytes) but not monkeys (Macaca mulatta).

Being able to recognize the faces of our friends and family members no matter where we see them represents a substantial challenge for the visual system because the retinal image of a face can be degraded by both changes in the person (age, expression, pose, hairstyle, etc.) and changes in the viewing conditions (direction and degree of illumination). Yet most of us are able to recognize familiar people effortlessly. A popular theory for how face recognition is achieved has argued that the brain stabilizes facial appearance by building average representations that enhance diagnostic features that reliably vary between people while diluting features that vary between instances of the same person. This explains why people find it easier to recognize average images of people, created by averaging multiple images of the same person together, than single instances (i.e. photographs). Although this theory is gathering momentum in the psychological and computer sciences, there is no evidence of whether this mechanism represents a unique specialization for individual recognition in humans. Here we tested two species, chimpanzees (Pan troglodytes) and rhesus monkeys (Macaca mulatta), to determine whether average images of different familiar individuals were easier to discriminate than photographs of familiar individuals. Using a two-alternative forced-choice, match-to-sample procedure, we report a behaviour response profile that suggests chimpanzees encode the faces of conspecifics differently than rhesus monkeys and in a manner similar to humans.

Familiar face + novel face = familiar face? Representational bias in the perception of morphed faces in chimpanzees

Highly social animals possess a well-developed ability to distinguish the faces of familiar from novel conspecifics to induce distinct behaviors for maintaining society. However, the behaviors of animals when they encounter ambiguous faces of familiar yet novel conspecifics, e.g., strangers with faces resembling known individuals, have not been well characterised. Using a morphing technique and preferential-looking paradigm, we address this question via the chimpanzee’s facial–recognition abilities. We presented eight subjects with three types of stimuli: (1) familiar faces, (2) novel faces and (3) intermediate morphed faces that were 50% familiar and 50% novel faces of conspecifics. We found that chimpanzees spent more time looking at novel faces and scanned novel faces more extensively than familiar or intermediate faces. Interestingly, chimpanzees looked at intermediate faces in a manner similar to familiar faces with regards to the fixation duration, fixation count, and saccade length for facial scanning, even though the participant was encountering the intermediate faces for the first time. We excluded the possibility that subjects merely detected and avoided traces of morphing in the intermediate faces. These findings suggest a bias for a feeling-of-familiarity that chimpanzees perceive familiarity with an intermediate face by detecting traces of a known individual, as 50% alternation is sufficient to perceive familiarity.

Beauty is in the ease of the beholding: a neurophysiological test of the averageness theory of facial attractiveness.

Hundreds of studies have shown that people prefer attractive over unattractive faces. But what is an attractive face, and why is it preferred? Averageness theory claims that faces are perceived as being attractive when their facial configuration approximates the mathematical average facial configuration of the population. Conversely, faces that deviate from this average configuration are perceived as being unattractive. The theory predicts that both attractive and mathematically averaged faces should be processed more fluently than unattractive faces, whereas the averaged faces should be processed marginally more fluently than the attractive faces. We compared neurocognitive and behavioral responses to attractive, unattractive, and averaged human faces to test these predictions. We recorded event-related potentials (ERPs) and reaction times (RTs) from 48 adults while they discriminated between human and chimpanzee faces. The participants categorized averaged and high-attractive faces as being "human" faster than low-attractive faces. The posterior N170 (150-225ms) face-evoked ERP component was smaller in response to high-attractive and averaged faces than to low-attractive faces. Single-trial electroencephalographic analysis indicated that this reduced ERP response arose from the engagement of fewer neural resources, and not from a change in the temporal consistency of how those resources were engaged. These findings provide novel evidence that faces are perceived as being attractive when they approximate a facial configuration close to the population average, and they suggest that processing fluency underlies preferences for attractive faces.

Neural representation of face familiarity in an awake chimpanzee

Evaluating the familiarity of faces is critical for social animals as it is the basis of individual recognition. In the present study, we examined how face familiarity is reflected in neural activities in our closest living relative, the chimpanzee. Skin-surface event-related brain potentials (ERPs) were measured while a fully awake chimpanzee observed photographs of familiar and unfamiliar chimpanzee faces (Experiment 1) and human faces (Experiment 2). The ERPs evoked by chimpanzee faces differentiated unfamiliar individuals from familiar ones around midline areas centered on vertex sites at approximately 200 ms after the stimulus onset. In addition, the ERP response to the image of the subject’s own face did not significantly diverge from those evoked by familiar chimpanzees, suggesting that the subject’s brain at a minimum remembered the image of her own face. The ERPs evoked by human faces were not influenced by the familiarity of target individuals. These results indicate that chimpanzee neural representations are more sensitive to the familiarity of conspecific than allospecific faces.

Laterality Effect for Faces in Chimpanzees (Pan troglodytes)

Face perception in humans is governed more by right-hemispheric than left-hemispheric neural correlate. Some but not all neurophysiological studies depict a right-side dominance for face responsive neurons in the brains of macaques. Hence, it is an open question whether and to what extent a right-hemisphere preference of processing faces exists across primate brains. We investigated chimpanzees discriminating chimeric faces of chimpanzees and humans, i.e., the combination of either left or right sides of a face vertically flipped and merged into a whole face. We found an effect of choosing the left-chimeric face more often than the right-chimeric face as being the one of the two that is closer to the original face, reflecting an advantage for the right side of the brain to process faces, as reported in humans. Moreover, we found a modulation by age of the participants, suggesting that the exposure history with a particular category shapes the right-hemispheric neural correlate to a configural/holistic processing strategy. In other words, the findings in chimpanzee participants parallel those in human participants and are suggestive for similar neural machineries in the occipital-temporal cortices in both species.

Developmental processes in face perception.

Understanding the developmental origins of face recognition has been the goal of many studies of various approaches. Contributions of experience-expectant mechanisms (early component), like perceptual narrowing, and lifetime experience (late component) to face processing remain elusive. By investigating captive chimpanzees of varying age, a rare case of a species with lifelong exposure to non-conspecific faces at distinctive levels of experience, we can disentangle developmental components in face recognition. We found an advantage in discriminating chimpanzee above human faces in young chimpanzees, reflecting a predominant contribution of an early component that drives the perceptual system towards the conspecific morphology, and an advantage for human above chimpanzee faces in old chimpanzees, reflecting a predominant late component that shapes the perceptual system along the critical dimensions of the face exposed to. We simulate the contribution of early and late components using computational modeling and mathematically describe the underlying functions.

The perception of two-tone Mooney faces in chimpanzees (Pan troglodytes)

Neurological experiments have revealed a complex network of areas in the human brain that respond more to faces than to other categories of objects and thus have been implemented in face categorization. The aim of this study was to investigate whether chimpanzees (n = 5), our closest living relatives, detect and categorize faces on the basis of first-order information, and whether this sensitivity is specific to faces or generalizes to other objects. In service to this aim, we created multiple categories of two-tone ‘Mooney’ objects (chimpanzee faces, shoes, human hands), because, by maximizing contrast, the Mooney transformation selectively degrades second-order information (the basis for individual discrimination in humans), leaving only first-order information intact. Two experiments used a 2AFC MTS procedure. The first experiment provided strong evidence that, like humans, chimpanzees categorize Mooney faces as faces. However, without second-order information, the chimpanzees could not match Mooney faces at the individual level. In Experiment 2, four of the five chimpanzees found it easier to categorize Mooney faces than Mooney shoes. Neurological evidence strongly indicates a dedicated neural mechanism for face categorization in the human brain, and our data suggest that chimpanzees share this level of specialization.

Evaluating factors of innate and experience-based contributions to chimpanzee face perception?

Evaluating factors of innate and experience-based contributions to chimpanzee face perception?

Chimpanzee Faces are ‘Special’ to Humans

A longstanding proposal is that primates, including humans, might have an innate representation of face structure. But, if humans have such a representation, how broad is its form: limited to coding conspecifics, or general enough to cover related species? The results reported here show adult humans process faces of chimpanzees in a way previously assumed to be exclusive to human faces. The composite effect was used to provide the first direct test of so-called holistic processing. Despite no lifetime experience of chimpanzees, adult humans showed a substantial composite effect for chimpanzee faces, and also an inversion effect. There was no similar evidence of holistic processing for faces of species of greater phylogenic distance from humans, including gorillas, spider monkeys, sheep, chickens, and Jacky lizards; nor was there any effect for non-face objects. In contrast to the holistic processing results, discrimination of chimpanzee faces was, as expected, poor. In the context of evidence that poor discrimination of heterospecific faces arises from a process of perceptual narrowing in infancy, our results suggest that adults retain some aspects of a broader bandwidth present in neonates (holistic processing) but lose others (discrimination).

Faces capture the visuospatial attention of chimpanzees (Pan troglodytes): evidence from a cueing experiment

BackgroundFaces, as socially relevant stimuli, readily capture human visuospatial attention. Although faces also play important roles in the social lives of chimpanzees, the closest living species to humans, the way in which faces are attentionally processed remains unclear from a comparative-cognitive perspective. In the present study, three young chimpanzees (Pan troglodytes) were tested with a simple manual response task in which various kinds of photographs, including faces as non-informative cues, were followed by a target.ResultsWhen the target appeared at the location that had been occupied by the face immediately before target onset, response times were significantly faster than when the target appeared at the opposite location that had been by the other object. Such an advantage was not observed when a photograph of a banana was paired with the other object. Furthermore, this attentional capture was also observed when upright human faces were presented, indicating that this effect is not limited to own-species faces. On the contrary, when the participants were tested with inverted chimpanzee faces, this effect was rather weakened, suggesting the specificity to upright faces.ConclusionChimpanzee's visuospatial attention was easily captured by the face stimuli. This effect was face specific and stronger for upright faces than inverted. These results are consistent with those from typically developing humans.