Robots in the service of animal behavior

Of rats and robots: A mutual learning paradigm

Robots are increasingly used alongside Skinner boxes to train animals in operant conditioning tasks. Similarly, animals are being employed in artificial intelligence research to train various algorithms. However, both types of experiments rely on unidirectional learning, where one partner-the animal or the robot-acts as the teacher and the other as the student. Here, we present a novel animal-robot interaction paradigm that enables bidirectional, or mutual, learning between a Wistar rat and a robot. The two agents interacted with each other to achieve specific goals, dynamically adjusting their actions based on the positive (rewarding) or negative (punishing) signals provided by their partner. The paradigm was tested in silico with two artificial reinforcement learning agents and in vivo with different rat-robot pairs. In the virtual trials, both agents were able to adapt their behavior toward reward maximization, achieving mutual learning. The in vivo experiments revealed that rats rapidly acquired the behaviors necessary to receive the reward and exhibited passive avoidance learning for negative signals when the robot displayed a steep learning curve. The developed paradigm can be used in various animal-machine interactions to test the efficacy of different learning rules and reinforcement schedules.

AI and the Doctor Dolittle challenge

Talking to animals is a fundamental human desire. The emergence of powerful AI algorithms, and specifically Large Language Models, has driven many to suggest that we are on the verge of fulfilling this wish. A few large scientific consortia have been formed around this topic and several commercial entities even offer such services. We frame the task of communicating with animals as 'The Doctor Dolittle challenge' and identify three main obstacles on the route to doing so. First, although generative AI models can create novel animal communication samples, it is very difficult to determine their context, and we will forever be biased by our human umwelt when doing so. Second, using AI to extract context in an unsupervised manner must be validated through controlled experiments aiming to measure the animals' response. This is difficult, and moreover, AI algorithms tend to cling on to any available information and are thus prone to finding spurious correlations. And third, animal communication focuses on a restricted set of contexts, such as alarm and courtship, highly limiting our ability to communicate regarding other contexts. Nevertheless, using the tremendous power of novel AI methods to decipher and mimic animal communication is both fascinating and important. We thus define the criteria for passing the Doctor Dolittle challenge and call upon scientists to take on the mission.

Influence of reward and location on dogs' behaviour toward an interactive artificial agent

Animal-robot interaction studies provide outstanding opportunities to understand the principles of social interactions. Here we investigated whether dogs' behaviour toward a cooperative artificial agent (Unidentified Moving Object (UMO)) is influenced by receiving a reward directly from the agent, and by variability in the UMO's location. In a problem-solving task, the UMO either helped dogs to obtain food (Direct Reward Group, DRG) or to fetch an object followed by an indirect reward from the owner/experimenter (Indirect Reward Group, IRG). During the Familiarization, the UMO either started from the same location or changed its starting location in all trials. In the Test phase, dogs faced the same task, but additionally a second, unfamiliar UMO was present. We found that both reward groups gazed at the UMO with decreasing latency during the Familiarization, with the IRG showing more gaze alternations between UMO and hiding location. Dogs showed no preference for either UMO in the Test phase but looked at the familiar UMO sooner if it had changed its location during the Familiarization. Thus, direct reward is not necessary to elicit elements of socially competent behavior in dogs, but variability in its motion may be important to improve the UMO's animacy and promote flexible learning.

Driving singing behaviour in songbirds using a multi-modal, multi-agent virtual environment

Interactive biorobotics provides unique experimental potential to study the mechanisms underlying social communication but is limited by our ability to build expressive robots that exhibit the complex behaviours of birds and small mammals. An alternative to physical robots is to use virtual environments. Here, we designed and built a modular, audio-visual 2D virtual environment that allows multi-modal, multi-agent interaction to study mechanisms underlying social communication. The strength of the system is an implementation based on event processing that allows for complex computation. We tested this system in songbirds, which provide an exceptionally powerful and tractable model system to study social communication. We show that pair-bonded zebra finches (Taeniopygia guttata) communicating through the virtual environment exhibit normal call timing behaviour, males sing female directed song and both males and females display high-intensity courtship behaviours to their mates. These results suggest that the environment provided is sufficiently natural to elicit these behavioral responses. Furthermore, as an example of complex behavioral annotation, we developed a fully unsupervised song motif detector and used it to manipulate the virtual social environment of male zebra finches based on the number of motifs sung. Our virtual environment represents a first step in real-time automatic behaviour annotation and animal–computer interaction using higher level behaviours such as song. Our unsupervised acoustic analysis eliminates the need for annotated training data thus reducing labour investment and experimenter bias.

Where Is Ethology Heading? An Invitation for Collective Metadisciplinary Discussion

Many factors can impact the advancement of scientific disciplines. In the study of animal behavior and cognition (i.e., Ethology), a lack of consensus about definitions or the emergence of some current events and inventions, among other aspects, may challenge the discipline's grounds within the next decades. A collective metadisciplinary discussion may help in envisioning the future to come. For that purpose, I elaborated an online questionnaire about the level of consensus and the researchers' ways of doing in seven areas: Discipline name and concepts, species, Umwelt, technology, data, networking, and the impact of sociocultural and ecological factors. I recruited the opinion of almost a hundred of colleagues worldwide (N = 98), both junior and seniors, working both in the wild and in the lab. While the results were pitted against the literature, general conclusions should be taken with caution and considered as a first attempt in exploring the state of the discipline from the researchers' perspective: There is no unanimity for the discipline's name; 71.4% of the researchers reported there is limited consensus in the definition of relevant concepts (i.e., culture, cognition); primate species still predominate in publications whereas the species selection criteria is sometimes based on fascination, chance, or funding opportunities rather than on biocentric questions; 56.1% of the apparatuses employed do not resemble species' ecological problems, and current tech needs would be solved by fostering collaboration with engineers. Finally, embracing the Open Science paradigm, supporting networking efforts, and promoting diversity in research teams may help in gathering further knowledge in the area. Some suggestions are proposed to overcome the aforementioned problems in this contemporary analysis of our discipline.

Locomotion of Miniature Soft Robots

Miniature soft robots are mobile devices, which are made of smart materials that can be actuated by external stimuli to realize their desired functionalities. Here, the key advancements and challenges of the locomotion producible by miniature soft robots in micro- to centimeter length scales are highlighted. It is highly desirable to endow these small machines with dexterous locomotive gaits as it enables them to easily access highly confined and enclosed spaces via a noninvasive manner. If miniature soft robots are able to capitalize this unique ability, they will have the potential to transform a vast range of applications, including but not limited to, minimally invasive medical treatments, lab-on-chip applications, and search-and-rescue missions. The gaits of miniature soft robots are categorized into terrestrial, aquatic, and aerial locomotion. Except for the centimeter-scale robots that can perform aerial locomotion, the discussions in this report are centered around soft robots that are in the micro- to millimeter length scales. Under each category of locomotion, prospective methods and strategies that can improve their gait performances are also discussed. This report provides critical analyses and discussions that can inspire future strategies to make miniature soft robots significantly more agile.

Model-based feedback control of live zebrafish behavior via interaction with a robotic replica

The possibility of regulating the behavior of live animals using biologically-inspired robots has attracted the interest of biologists and engineers for over twenty-five years. From early work on insects to recent endeavors on mammals, we have witnessed fascinating applications that have pushed forward our understanding of animal behavior along new directions. Despite significant progress, most of the research has focused on open-loop control systems, in which robots execute predetermined actions, independent of the animal behavior. We integrate mathematical modeling of social behavior toward the design of realistic feedback laws for robots to interact with a live animal. In particular, we leverage recent advancements in data-driven modeling of zebrafish behavior. Ultimately, we establish a novel robotic platform that allows real-time actuation of a biologically-inspired 3D-printed zebrafish replica to implement model-based control of animal behavior. We demonstrate our approach through a series of experiments, designed to elucidate the appraisal of the replica by live subjects with respect to conspecifics and to quantify the biological value of closed-loop control.

Socialization in Commercial Breeding Kennels: The Use of Novel Stimuli to Measure Social and Non-Social Fear in Dogs

Simple Summary

Owner-based reports of dogs presumed to come from commercial breeding kennels (CBKs) suggest high levels of fear in this population. Fear in kenneled dogs is a serious behavioral welfare concern as it may lead to both acute and chronic stress. Novel social and non-social stimuli have been shown to elicit behaviors associated with fear in animals. New knowledge on the levels of fear in dogs from CBKs could be used to further refine protocols intended for assessment of welfare in CBKs and to improve breeders’ management practices. The aim of this study, therefore, was to evaluate how dogs from CBKs reacted to social (i.e., a person approaching) and non-social (i.e., a traffic cone and a dog statue) stimuli, and to perform a preliminary evaluation of how these responses might be used as indicators of dogs’ overall socialization levels in kennels. Results revealed that dogs had primarily mildly fearful responses to the stimuli presented. These findings are encouraging as extreme fearful reactions were rarely recorded. Nevertheless, there is a clear margin for commercial breeders to improve the socialization protocols in their kennels to better incorporate both social and non-social stimuli.

Abstract

Understanding the behavioral welfare of dogs in commercial breeding kennels (CBKs) is important for improving breeders’ management practices as well as dog welfare. In the current study, breeding dogs from CBKs were exposed to novel stimuli to evaluate their behavioral responses, with emphasis on indicators of fear. Subjects were presented with a standard stranger-approach test, a traffic cone, and a realistic dog statue. Sixty dogs were exposed to the three stimuli and behavioral responses were scored using an ethogram developed for this study. Dogs spent significantly more time investigating the environment, staying further away from the stimulus, and they took longer to approach and investigate when presented with the cone than with the dog statue or stranger (p < 0.01). These findings suggest that the cone elicited more fear-related behaviors than the dog statue and stranger. Given these results, in addition to socializing their dogs to unfamiliar people and other dogs within their kennels, commercial breeders should be encouraged to increase the exposure of their dogs to more diverse novel stimuli to reduce non-social fear and support the welfare of dogs while they reside in the kennel and when they transition to new homes.

Navigation of a Freely Walking Fruit Fly in Infinite Space Using a Transparent Omnidirectional Locomotion Compensator (TOLC)

Animal behavior is an essential element in behavioral neuroscience study. However, most behavior studies in small animals such as fruit flies (Drosophilamelanogaster) have been performed in a limited spatial chamber or by tethering the fly's body on a fixture, which restricts its natural behavior. In this paper, we developed the Transparent Omnidirectional Locomotion Compensator (TOLC) for a freely walking fruit fly without tethering, which enables its navigation in infinite space. The TOLC maintains a position of a fruit fly by compensating its motion using the transparent sphere. The TOLC is capable of maintaining the position error < 1 mm for 90.3% of the time and the heading error < 5° for 80.2% of the time. The inverted imaging system with a transparent sphere secures the space for an additional experimental apparatus. Because the proposed TOLC allows us to observe a freely walking fly without physical tethering, there is no potential injury during the experiment. Thus, the TOLC will offer a unique opportunity to investigate longitudinal studies of a wide range of behavior in an unrestricted walking Drosophila.

Complex sensory environments alter mate choice outcomes

Noise is a common problem in animal communication. We know little, however, about how animals communicate in the presence of noise using multimodal signals. Multimodal signals are hypothesised to be favoured by evolution because they increase the efficacy of detection and discrimination in noisy environments. We tested the hypothesis that female túngara frogs' responses to attractive male advertisement calls are improved in noise when a visual signal component is added to the available choices. We tested this at two levels of decision complexity (two and three choices). In a two-choice test, the presence of noise did not reduce female preferences for attractive calls. The visual component of a calling male, associated with an unattractive call, also did not reduce preference for attractive calls in the absence of noise. In the presence of noise, however, females were more likely to choose an unattractive call coupled with the visual component. In three-choice tests, the presence of noise alone reduced female responses to attractive calls and this was not strongly affected by the presence or absence of visual components. The responses in these experiments fail to support the multimodal signal efficacy hypothesis. Instead, the data suggest that audio-visual perception and cognitive processing, related to mate choice decisions, are dependent on the complexity of the sensory scene.

Sensory ecology of the frog-eating bat, Trachops cirrhosus, from DNA metabarcoding and behavior

Metabarcoding of prey DNA from fecal samples can be used to design behavioral experiments to study the foraging behavior and sensory ecology of predators. The frog-eating bat, Trachops cirrhosus, eavesdrops on the mating calls of its anuran prey. We captured wild T. cirrhosus and identified prey remains in the bats’ fecal samples using DNA metabarcoding of two gene regions (CO1 and 16S). Bats were preying on frogs previously unknown in their diet, such as species in the genus Pristimantis, which occurred in 29% of T. cirrhosus samples. Twenty-three percent of samples also contained DNA of Anolis lizards. We additionally report apparently rare predation events on hummingbirds and heterospecific bats. We used results from metabarcoding to design acoustic and 3D model stimuli to present to bats in behavioral experiments. We show predatory responses by T. cirrhosus to the calls of the frog Pristimantis taeniatus and to the rustling sounds of anoles moving through leaf-litter, as well as attacks on a stuffed hummingbird and a plastic anole model. The combination of species-specific dietary information from metabarcoding analyses with behavioral responses to prey cues provides a unique window into the foraging ecology of predators that are difficult to observe in the wild.

Animals in Virtual Environments

The core idea in an XR (VR/MR/AR) application is to digitally stimulate one or more sensory systems (e.g. visual, auditory, olfactory) of the human user in an interactive way to achieve an immersive experience. Since the early 2000s biologists have been using Virtual Environments (VE) to investigate the mechanisms of behavior in non-human animals including insects, fish, and mammals. VEs have become reliable tools for studying vision, cognition, and sensory-motor control in animals. In turn, the knowledge gained from studying such behaviors can be harnessed by researchers designing biologically inspired robots, smart sensors, and rnulti-agent artificial intelligence. VE for animals is becoming a widely used application of XR technology but such applications have not previously been reported in the technical literature related to XR. Biologists and computer scientists can benefit greatly from deepening interdisciplinary research in this emerging field and together we can develop new methods for conducting fundamental research in behavioral sciences and engineering. To support our argument we present this review which provides an overview of animal behavior experiments conducted in virtual environments.

Perceived Synchrony of Frog Multimodal Signal Components Is Influenced by Content and Order

Multimodal signaling is common in communication systems. Depending on the species, individual signal components may be produced synchronously as a result of physiological constraint (fixed) or each component may be produced independently (fluid) in time. For animals that rely on fixed signals, a basic prediction is that asynchrony between the components should degrade the perception of signal salience, reducing receiver response. Male túngara frogs, Physalaemus pustulosus, produce a fixed multisensory courtship signal by vocalizing with two call components (whines and chucks) and inflating a vocal sac (visual component). Using a robotic frog, we tested female responses to variation in the temporal arrangement between acoustic and visual components. When the visual component lagged a complex call (whine + chuck), females largely rejected this asynchronous multisensory signal in favor of the complex call absent the visual cue. When the chuck component was removed from one call, but the robofrog inflation lagged the complex call, females responded strongly to the asynchronous multimodal signal. When the chuck component was removed from both calls, females reversed preference and responded positively to the asynchronous multisensory signal. When the visual component preceded the call, females responded as often to the multimodal signal as to the call alone. These data show that asynchrony of a normally fixed signal does reduce receiver responsiveness. The magnitude and overall response, however, depend on specific temporal interactions between the acoustic and visual components. The sensitivity of túngara frogs to lagging visual cues, but not leading ones, and the influence of acoustic signal content on the perception of visual asynchrony is similar to those reported in human psychophysics literature. Virtually all acoustically communicating animals must conduct auditory scene analyses and identify the source of signals. Our data suggest that some basic audiovisual neural integration processes may be at work in the vertebrate brain.

Closed-loop control of zebrafish behaviour in three dimensions using a robotic stimulus

Robotics is continuously being integrated in animal behaviour studies to create customizable, controllable, and repeatable stimuli. However, few systems have capitalized on recent breakthroughs in computer vision and real-time control to enable a two-way interaction between the animal and the robot. Here, we present a "closed-loop control" system to investigate the behaviour of zebrafish, a popular animal model in preclinical studies. The system allows for actuating a biologically-inspired 3D-printed replica in a 3D workspace, in response to the behaviour of a zebrafish. We demonstrate the role of closed-loop control in modulating the response of zebrafish, across a range of behavioural and information-theoretic measures. Our results suggest that closed-loop control could enhance the degree of biomimicry of the replica, by increasing the attraction of live subjects and their interaction with the stimulus. Interactive experiments hold promise to advance our understanding of zebrafish, offering new means for high throughput behavioural phenotyping.

Multiple cues produced by a robotic fish modulate aggressive behaviour in Siamese fighting fishes

The use of robotics to establish social interactions between animals and robots, represents an elegant and innovative method to investigate animal behaviour. However, robots are still underused to investigate high complex and flexible behaviours, such as aggression. Here, Betta splendens was tested as model system to shed light on the effect of a robotic fish eliciting aggression. We evaluated how multiple signal systems, including a light stimulus, affect aggressive responses in B. splendens. Furthermore, we conducted experiments to estimate if aggressive responses were triggered by the biomimetic shape of fish replica, or whether any intruder object was effective as well. Male fishes showed longer and higher aggressive displays as puzzled stimuli from the fish replica increased. When the fish replica emitted its full sequence of cues, the intensity of aggression exceeded even that produced by real fish opponents. Fish replica shape was necessary for conspecific opponent perception, evoking significant aggressive responses. Overall, this study highlights that the efficacy of an artificial opponent eliciting aggressive behaviour in fish can be boosted by exposure to multiple signals. Optimizing the cue combination delivered by the robotic fish replica may be helpful to predict escalating levels of aggression.

Information theory and robotics meet to study predator-prey interactions

Transfer entropy holds promise to advance our understanding of animal behavior, by affording the identification of causal relationships that underlie animal interactions. A critical step toward the reliable implementation of this powerful information-theoretic concept entails the design of experiments in which causal relationships could be systematically controlled. Here, we put forward a robotics-based experimental approach to test the validity of transfer entropy in the study of predator-prey interactions. We investigate the behavioral response of zebrafish to a fear-evoking robotic stimulus, designed after the morpho-physiology of the red tiger oscar and actuated along preprogrammed trajectories. From the time series of the positions of the zebrafish and the robotic stimulus, we demonstrate that transfer entropy correctly identifies the influence of the stimulus on the focal subject. Building on this evidence, we apply transfer entropy to study the interactions between zebrafish and a live red tiger oscar. The analysis of transfer entropy reveals a change in the direction of the information flow, suggesting a mutual influence between the predator and the prey, where the predator adapts its strategy as a function of the movement of the prey, which, in turn, adjusts its escape as a function of the predator motion. Through the integration of information theory and robotics, this study posits a new approach to study predator-prey interactions in freshwater fish.

What artifice can and cannot tell us about animal behavior

Artifice-the manipulation of social and environmental stimuli-is fundamental to research in animal behavior. State-of-the-art techniques have been developed to generate and present complex visual stimuli. These techniques have unique strengths and limitations. However, many of the issues with synthetic animation and virtual reality are common to playback experiments in general, including those using unmanipulated video or auditory stimuli. Playback experiments, in turn, fall into the broader category of experiments that artificially manipulate the array of stimuli experienced by a subject. We argue that the challenges of designing and interpreting experiments using virtual reality or synthetic animations are largely comparable to those of studies using older technologies or addressing other modalities, and that technology alone is unlikely to solve these challenges. We suggest that appropriate experimental designs are the key to validating behavioral responses to artificial stimuli and to interpreting all studies using artifice, including those that present complex visual displays.

The anuran vocal sac: a tool for multimodal signalling

Although in anurans the predominant mode of intra- and intersexual communication is vocalization, modalities used in addition to or instead of acoustic signals range from seismic and visual to chemical. In some cases, signals of more than one modality are produced through or by the anuran vocal sac. However, its role beyond acoustics has been neglected for some time and nonacoustic cues such as vocal sac movement have traditionally been seen as an epiphenomenon of sound production. The diversity in vocal sac coloration and shape found in different species is striking and recently its visual properties have been given a more important role in signalling. Chemosignals seem to be the dominant communication mode in newts, salamanders and caecilians and certainly play a role in the aquatic life phase of anurans, but airborne chemical signalling has received less attention. There is, however, increasing evidence that at least some terrestrial anuran species integrate acoustic, visual and chemical cues in species recognition and mate choice and a few secondarily mute anuran species seem to fully rely on volatile chemical cues produced in glands on the vocal sac. Within vertebrates, frogs in particular are suitable organisms for investigating multimodal communication by means of experiments, since they are tolerant of disturbance by observers and can be easily manipulated under natural conditions. Thus, the anuran vocal sac might be of great interest not only to herpetologists, but also to behavioural biologists studying communication systems.

•

Our view of anuran communication has changed drastically during the last few decades.

•

Anurans use diverse communication strategies often directly related to the vocal sac.

•

Distinctive vocal sac shapes, sizes and colours suggest functional diversity.

•

Evidence increases that the vocal sac also plays a role in chemical signalling.

•

The anuran vocal sac may be of great interest for studies on communication systems.

From uni- to multimodality: towards an integrative view on anuran communication

Undeniably, acoustic signals are the predominant mode of communication in frogs and toads. Acoustically active species are found throughout the vast diversity of anuran families. However, additional or alternative signal modalities have gained increasing attention. In several anurans, seismic, visual and chemical communications have convergently evolved due to ecological constraints such as noisy environments. The production of a visual cue, like the inevitably moving vocal sac of acoustically advertising males, is emphasized by conspicuously coloured throats. Limb movements accompanied by dynamic displays of bright colours are additional examples of striking visual signals independent of vocalizations. In some multimodal anuran communication systems, the acoustic component acts as an alert signal, which alters the receiver attention to the following visual display. Recent findings of colourful glands on vocal sacs, producing volatile species-specific scent bouquets suggest the possibility of integration of acoustic, visual and chemical cues in species recognition and mate choice. The combination of signal components facilitates a broadened display repertoire in challenging environmental conditions. Thus, the complexity of the communication systems of frogs and toads may have been underestimated.