Bernd Meyer

A bstract: Division of labour is fundamental to the functioning of societies and socially living organisms. While it has been central to their study for decades, no complete picture has emerged yet. Some of the most fascinating questions arise in the context of self-organised societies, like those of social insects, that coordinate their behaviour with completely decentralised simple decision-making performed by individuals that only have local information at their disposal. Based on empirical evidence, these collectives appear to balance task engagement globally across their whole task network for the benefit of the colony overall. How can this pro-social coordination be achieved by independently acting individuals? How is a global workforce balancing possible based on only local perception with no knowledge of the global colony status or needs? Central to solving these problems is the question how can the relevant information flow through the task network so that a changed task demand in one part of the colony can lead to adjustments in distant other parts? We detail a model that presents a potential answer to this conundrum. Our model is informed by evolutionary game theory and rests on the assumption that the perception of an individual’s sensory input can evolve. We present simulation studies and a mathematical proof to show that pro-social behaviour will evolve in a collective of agents that adjust their behaviour using primitive and biologically plausible learning mechanisms if we assume an evolving perception function.

The scarcity of labelled data makes training Deep Neural Network (DNN) models in bioacoustic applications challenging. In typical bioacoustics applications, manually labelling the required amount of data can be prohibitively expensive. To effectively identify both new and current classes, DNN models must continue to learn new features from a modest amount of fresh data. Active Learning (AL) is an approach that can help with this learning while requiring little labelling effort. Nevertheless, the use of fixed feature extraction approaches limits feature quality, resulting in underutilization of the benefits of AL. We describe an AL framework that addresses this issue by incorporating feature extraction into the AL loop and refining the feature extractor after each round of manual annotation. In addition, we use raw audio processing rather than spectrograms, which is a novel approach. Experiments reveal that the proposed AL framework requires 14.3%, 66.7%, and 47.4% less labelling effort on benchmark audio datasets ESC-50, UrbanSound8k, and InsectWingBeat, respectively, for a large DNN model and similar savings on a microcontroller-based counterpart. Furthermore, we showcase the practical relevance of our study by incorporating data from conservation biology projects. All codes are publicly available on GitHub.

Tracking individuals is a vital part of many experiments conducted to understand collective behaviour. Ants are the paradigmatic model system for such experiments but their lack of individually distinguishing visual features and their high colony densities make it extremely difficult to perform reliable racking automatically. Additionally, the wide diversity of their species' appearances makes a generalized approach even harder. In this paper, we propose a data-driven multi-object tracker that, for the first time, employs domain adaptation to achieve the required generalisation. This approach is built upon a joint-detection-and-tracking framework that is extended by a set of domain discriminator modules integrating an adversarial training strategy in addition to the tracking loss. In addition to this novel domain-adaptive tracking framework, we present a new dataset and a benchmark for the ant tracking problem. The dataset contains 57 video sequences with full trajectory annotation, including 30k frames captured from two different ant species moving on different background patterns. It comprises 33 and 24 sequences for source and target domains, respectively. We compare our proposed framework against other domain-adaptive and non-domain-adaptive multi-object tracking baselines using this dataset and show that incorporating domain adaptation at multiple levels of the tracking pipeline yields significant improvements. The code and the dataset are available at https://github.com/chamathabeysinghe/da-tracker.

Division of labour, or the differentiation of the individuals in a collective across tasks, is a fundamental aspect of social organisations, such as social insect colonies. It allows for efficient resource use and improves the chances of survival for the entire collective. The emergence of large inactive groups of individuals in insect colonies sometimes referred to as laziness, has been a puzzling and hotly debated division-of-labour phenomenon in recent years that is counter to the intuitive notion of effectiveness. It has previously been shown that inactivity can be explained as a by-product of social learning without the need to invoke an adaptive function. While highlighting an interesting and important possibility, this explanation is limited because it is not yet clear whether the relevant aspects of colony life are governed by social learning. In this paper, we explore the two fundamental types of behavioural adaptation that can lead to a division of labour, individual learning and social learning. We find that inactivity can just as well emerge from individual learning alone. We compare the behavioural dynamics in various environmental settings under the social and individual learning assumptions, respectively. We present individual-based simulations backed up by analytic theory, focusing on adaptive dynamics for the social paradigm and cross-learning for the individual paradigm. We find that individual learning can induce the same behavioural patterns previously observed for social learning. This is important for the study of the collective behaviour of social insects because individual learning is a firmly established paradigm of behaviour learning in their colonies. Beyond the study of inactivity, in particular, the insight that both modes of learning can lead to the same patterns of behaviour opens new pathways to approach the study of emergent patterns of collective behaviour from a more generalised perspective.

Tracking individuals is a vital part of many experiments conducted to understand collective behaviour. Ants are the paradigmatic model system for such experiments but their lack of individually distinguishing visual features and their high colony densities make it extremely difficult to perform reliable tracking automatically. Additionally, the wide diversity of their species' appearances makes a generalized approach even harder. In this paper, we propose a data-driven multi-object tracker that, for the first time, employs domain adaptation to achieve the required generalisation. This approach is built upon a joint-detection-and-tracking framework that is extended by a set of domain discriminator modules integrating an adversarial training strategy in addition to the tracking loss. In addition to this novel domain-adaptive tracking framework, we present a new dataset and a benchmark for the ant tracking problem. The dataset contains 57 video sequences with full trajectory annotation, including 30k frames captured from two different ant species moving on different background patterns. It comprises 33 and 24 sequences for source and target domains, respectively. We compare our proposed framework against other domain-adaptive and non-domain-adaptive multi-object tracking baselines using this dataset and show that incorporating domain adaptation at multiple levels of the tracking pipeline yields significant improvements. The code and the dataset are available at https://github.com/chamathabeysinghe/da-tracker.

Social insects are among the ecologically most successful collectively living organisms, with efficient division of labour a key feature of this success. Surprisingly, these efficient colonies often have a large proportion of inactive workers in their workforce, sometimes referred to aslazy workers. The dominant hypotheses explaining this are based on specific life-history traits, specific behavioural features or uncertain environments where inactive workers can provide a ‘reserve’ workforce that can spring into action quickly. While there is a number of experimental studies that show and investigate the presence of inactive workers, mathematical and computational models exploring specific hypotheses are not common. Here, using a simple mathematical model, we show that a parsimonious hypothesis can explain this puzzling social phenomenon. Our model incorporates social interactions and environmental influences into a game-theoretical framework and captures how individuals react to environment by allocating their activity according to environmental conditions. This model shows that inactivity can emerge under specific environmental conditions as a by-product of the task allocation process. Our model confirms the empirical observation that in the case of worker loss, prior homeostatic balance is re-established by replacing some of the lost force with previously inactive workers. Most importantly, our model shows that inactivity in social colonies can be explained without the need to assume an adaptive function for this phenomenon.

Molecules consisting of a donor and an acceptor moiety can exhibit large intrinsic dipole moments. Upon deposition on a metal surface, the dipole may be effectively screened and the charge distribution altered due to hybridization with substrate electronic states. Here, we deposit Ethyl-Diaminodicyanoquinone molecules, which exhibit a large dipole moment in gas phase, on a Au(111) surface. Employing a combination of scanning tunneling microscopy and non-contact atomic force microscopy, we find that a significant dipole moment persists in the flat-lying molecules. Density-functional theory calculations reveal that the dipole moment is even increased on the metal substrate as compared to the gas phase. We also show that the local contact potential across the molecular islands is decreased by several tens of meV with respect to the bare metal. We explain this by the induced charge-density redistribution due to the adsorbed molecules, which confine the substrate's wavefunction at the interface. Our local measurements provide direct evidence of this so-called push-back or cushion effect at the scale of individual molecules.