We present a novel artificial cognitive map system using the generative deep neural networks called Variational Autoencoder / Generative Adversarial Network (VAE/GAN), which encodes input images into the latent space and the structure of the latent space is self-organized through the learning. Our results show that the distance of the predicted image is reflected in the distance of the corresponding latent vector after training, which indicates that the latent space is organized to reflect the proximity structure of the dataset. This system is also able to internally generate temporal sequences analogous to hippocampal replay/pre-play, and we found that these sequences are not just the exact replay of the past experience, and this could be the origin of creating novel sequences from the past experiences. Having this generative nature of cognition is thought as a prerequisite for artificial cognitive systems.

In this study, we report the investigations conducted on the mimetic behavior of a new humanoid robot called Alter3. Alter3 autonomously imitates the motions of a human in front of him and stores the motion sequences in its memory. Alter3 also contains a self-simulator that simulates its own motions before executing them and generates a self-image. We investigate how this mimetic behavior evolves with human interaction, by analyzing memory dynamics and information flow between Alter3 and humans. One important observation from this study is that when Alter3 fails to imitate human motion, humans tend to imitate Alter3 instead. This tendency is quantified by the alternation of the direction of information flow. At the conference we will also report on the experiments we carried out recently, in which two Alters imitated each other, and in which we let people possess and imitate Alter.

The emergence of novel forms in evolution can be viewed as the exploration of new regions in evolutionary space. This study investigates exploratory dynamics in evolutionary spaces through the empirical analysis of a social tagging system, which considers tags as an evolving entity and the tag set space as an evolutionary space. Dimensionality reduction showed distribution of a tag set in high dimensional tag set space embedded in 2-dimensional space and suggested that the new use of common tags was explored around common use, while new use of an uncommon tag was explored multi-regionally. Exploratory paths of evolution in tag set space were visualized as directed networks, and they exhibited structures called “branch” and “bunch.” The former suggests exploration deep into the space and the latter indicates wide exploration. These two modes of exploration imply exploration dynamics in evolutionary space in a mining-like manner in which prospects deposit novel tag sets by digging deep and excavating a wide area of strata with deposits.

Social network services (SNSs) are examples of non-living systems that evolve in response to internal and external events and have many similar characteristics assumed in biological evolution. In the present study, we analyzed the evolution of hashtag use on an SNS called RoomClip. Using a biological evolution analogy, we viewed each post (photo submission) as a species and each set of associated hashtags with a photo as genome. Further, we virtually defined parent–offspring relationships among posts based on their hashtag use and observed the resulting family tree of posts. Our analysis revealed that there was weak selection on hashtag usages relative to the Yule–Simon processes with strong feedback, and hashtag use quickly diverged. The evolution of novel hashtag combinations was observed, which is more salient than an evolution of individual novel hashtags.

Research is being conducted to understand human social interactions on the Web as a biological ecosystem. Keystone species in a biological ecosystem are defined as species that significantly impact the ecosystem if removed, irrespective of its low biomass. Identifying keystone species is an important issue, as they play a vital role in maintaining the entire ecosystem and its biodiversity. We hypothesize that a Web system is akin to an open, living ecological system that evolves and sustains itself by constantly updating its elements, which are sustained by the emergence of keystone species. We use data from an online bulletin board and identify keystone threads (”species”) that have a large impact if they are removed or become unpopular, despite their small population size. Our analysis confirms that keystone threads do exist in the system. The system seems to asymptotically evolve to a critical state. At the same time, the number of keystone species increases, and metabolism is enhanced. From a network topological perspective, the system evolves into a network with a high degree, closeness, and PageRank centralities. These findings suggest that keystone species play an important role in the evolution of online ecosystems. Further, by having keystone species, the system itself can decrease stability and bring about diversity to the ecosystem; consequently, the system can evolve.

Acoustic ecologist Bernie Krause hypothesized that rich soundscapes in mature ecosystems are generated by sound communication between different species with differentiating acoustic niches. This hypothesis, called the acoustic niche hypothesis, proposes that in a mature ecosystem, the singing of a species occupies a unique bandwidth in frequency and shifts in time to avoid competition, thus making the communication efficient. We hypothesize that selective pressure on communication complexity is required for differentiating and filling acoustic niches by a limited number of species, in addition to selective pressures on communication efficiency. To test this hypothesis, we built an evolutionary model where agents can emit complex sounds. Our simulations with the model demonstrate that selective pressure on communication efficiency and complexity leads to an evolution in spectral differentiation with a limited number of species filling the acoustic niche. This is the first demonstration of acoustic niche differentiation using an artificial life model with complex-sounding agents. We also propose multi-timescale complexity measurement, extending the Jensen–Shannon complexity using multi-scale permutation entropy. We analyze the evolved soundscape in the simulations using this measure. The result shows that multi-timescale complexity in soundscape evolved, suggesting that evolving niche differentiation leads to ecological complexity. We implement the extended model in real space and demonstrate that the system can adaptively generate sounds, differentiating acoustic niches with environmental sounds.

Our previous study showed that embodied cultured neural networks and spiking neural networks with spike-timing dependent plasticity (STDP) can learn a behavior as they avoid stimulation from outside. In a sense, the embodied neural network can autonomously change their activity to avoid external stimuli. In this paper, as a result of our experiments using cultured neurons, we find that there is also a second property allowing the network to avoid stimulation: if the network cannot learn to avoid the external stimuli, it tends to decrease the stimulus-evoked spikes as if to ignore the input neurons. We also show such a behavior is reproduced by spiking neural networks with asymmetric-STDP. We consider that these properties can be regarded as autonomous regulation of self and non-self for the network.

While technology has brought immeasurable benefits to humankind, recent advances in artificial intelligence and autonomous systems have also led to new ethical, legal, and social issues. We now face the problem of creating a cooperative society in which autonomous systems and people can coexist. The concept of artificial life provides unique perspectives, tools, and philosophies for furthering our understanding of complex living, lifelike, or hybrid systems. However, artificial life is still difficult to comprehend for those outside the academic community. We thus created a public co-creation community called ALIFE Lab, which aims to increase awareness of artificial life in collaboration with artificial life researchers and talents from creative fields such as design, art, and fashion. As one of the community activities, we organized a workshop-based program in which participants learned about Artificial Life and used it as a tool to conceive autonomous systems with concrete vocabulary and theory. This paper reports the methodology and outcomes of the workshop.

Honeybees are highly social animals who live in large colonies, called hives. This study explores global patterns and dynamics observed in the beehive by tracking the individual behaviors. Previous research developed a high-throughput automatic monitoring system for honeybees ( Apis mellifera ) that tracked every individual bee in a hive, recording their positions, speed and orientations. This has been used to analyze the bees trophallaxis (two bees touching each other with their antennae to orally transferring liquid food (Free (1956))). network and calculate how often they communicate; it was found that the bee networks communicate in the intermittent manner in time, called bursts, much like human communication networks (Gernat et al. (2018)). Using this same dataset, we developed a new, complementary analysis that examined a different bee behavior that also follows a burst pattern: the bursts of kinetic energy that occur in beehives. Such bursts may be endogenous (i.e., spontaneous activity resulting from the internal interactions of bees) or exogenous (i.e., resulting from external perturbations). We sought to identify relationships between endogenous and exogenous bursts and the contributions of individual bees, as well as the relationship between the bees trophallaxis network and their kinetic bursting behaviors.

The complete Proceedings of the The 2018 Conference on Artificial Life: A Hybrid of the European Conference on Artificial Life (ECAL) and the International Conference on the Synthesis and Simulation of Living Systems (ALIFE)

Spiking neural networks with spike-timing dependent plasticity (STDP) can learn to avoid the external stimulations spontaneously. This principle is called "Learning by Stimulation Avoidance" (LSA) and can be used to reproduce learning experiments on cultured biological neural networks. LSA has promising potential, but its application and limitations have not be studied extensively. This paper focuses on the scalability of LSA for large networks and shows that LSA works well in small networks (100 neurons) and can be scaled to networks up to approximately 3,000 neurons.

Based on the principle of artificial life, we developed an upper-body android called "Alter." Alter is human-like in appearance, receives sensory information from the outside via an autonomous sensor system located around the android, and moves spontaneously using two autonomous systems of internal dynamics. Its body and arms contain a central pattern generator with seven degrees of freedom and hundreds of plastic artificial neurons. We investigated Alter’s environmental adaptability and the spontaneity of is behavioral patterns. In addition, we discuss the conditions under which a robot can become lifelike.

Collective behavior, swarming and mutual interactions between living entities are well-known phenomena in biology where, for example, flocks of birds or schools of fish are extensively studied. Swarming has also been observed in bacteria or tumor cell populations. This kind of collective behavior can be implemented also in non-living systems, namely in biologically-inspired swarm robotics. However, the collective behavior of chemical droplets (mutual interactions of multiple “liquid robots”) has not been studied before and the present paper reports the experimentally observed phenomena and modes of behavior in such system. We show how multiple decanol droplets in a thin layer of decanoate solution behave and interact. We report, for the first time, several life-like features such as collective chemotaxis and number-dependent group formation.

We propose the Epsilon Network (ε-network), a network that automatically adjusts its size to the complexity of a stream of data while performing online learning. The network optimises its topology during training, simultaneously adding and removing neurons and weights: it adds neurons where they can raise performance, and removes redundant neurons while preserving performance. The network is a neural realisation of the ε-machine devised by Crutchfield and al. (Crutchfield and Young (1989)). In this paper our network is trained to predict video frames; we evaluate it on simple, complex, and noisy videos and show that the final number of neurons is a good indicator of the complexity and predictability of the data stream.

Time perception is the capacity to sense the passing of time, but in most living creatures it also involves memorizing how much time passed, and eventually acting when it reaches a specific amount. The later is referred as interval timing. This capacity allows animals to detect temporally repeating events in their environment, avoid them if necessary, or exploit them if beneficial(Saigusa et al., 2008). While the research in animals has focused on interval timing (Connor, 1985; Durstewitz, 2003), research in artificial life has limited itself to time perception (Maniadakis et al., 2014; Trianni, 2008). Indeed, alife models rely on the intrinsic temporal properties of neural networks to encode the passing of time and, therefore, cannot estimate how much time passed since the onset of a stimulus. Our work attempts to make one step closer to interval timing by designing an agent which must learn the duration of a stimulus, but also replay it later on.

We present some arguments why existing methods for representing agents fall short in applications crucial to artificial life. Using a thought experiment involving a fictitious dynamical systems model of the biosphere we argue that the metabolism, motility, and the concept of counterfactual variation should be compatible with any agent representation in dynamical systems. We then propose an information-theoretic notion of integrated spatiotemporal patterns which we believe can serve as the basic building block of an agent definition. We argue that these patterns are capable of solving the problems mentioned before. We also test this in some preliminary experiments.