An understanding of analogy and the multiple realizability of concepts, ideas, and experience is necessary to understand cognition and the generation of behavior even at the most abstract levels. One of the most fundamental questions one can ask about a pair of neural circuits is whether they are doing the same thing or different things. Our work addresses this question by applying a model of sequential narrative analogy, Net-MATCH, to neural circuits evolved to perform a simple locomotion task. Along the way, we develop a measure of the “experience” of a neural circuit performing a behavior we call its functional trace. We find (i) that Net-MATCH reports strong analogies between some, but not all, neural circuits that perform the walking behavior, (ii) that it finds stronger analogies between circuits of the same class (as reported in previous work on this problem space) than circuits of different classes, and (iii) that it reveals strong analogies between circuits of the previously-reported BS-switch and SW-switch classes, even though these classes are of different circuit sizes. We conclude that Net-MATCH is a powerful tool for understanding the multiple realizability of behavior.

The major evolutionary transition to multicellularity shifted the unit of selection from individual cells to multicellular organisms. Constituent cells must regulate their growth and cooperate to benefit the whole organism, even when such behaviors would have been maladaptive were they free living. Mutations that disrupt cellular cooperation can lead to various ailments, including physical deformities and cancer. Organisms therefore employ mechanisms to enforce cooperation, such as error correction, policing, and genetic robustness. We built a simulation to study this last mechanism under a range of evolutionary conditions. Specifically, we asked: How does genetic robustness against cellular cheating evolve in multicellular organisms? We focused on early multicellular organisms (with only one cell type) where cells must control their growth to avoid overwriting each other. In our model, unrestrained cells will outcompete restrained cells within an organism, but restrained cells alone will result in faster reproduction for the organism. Ultimately, we demonstrate a clear selective pressure for genetic robustness in multicellular organisms and show that this pressure increases with the total number of cells in the organism.

Several studies that deal with the acquisition of concepts in a bottom-up manner from experiences in the physical space exist, but there are few of them that deal with the bidirectional interaction between symbolic operations and experiences in the physical world. It was shown that a shared module neural network succeeded in generating a bottom-up spatial representation of the external world, without involving learning of the signals of the spatial structure. Furthermore, the module can understand the external map as a symbol based on its spatial representation, and top-down navigation can be performed using the map. In this study, we extended this model and proposed a simulation model that unifies the emergence of a number representation, learning of symbol manipulation on the representation, and top-down understanding of symbol manipulation onto the physical world. Our results show that the learning results of the symbol manipulation can be applied to the physical world prediction, and our proposed model succeeded in grounding symbol manipulation onto physical experiences.

Endosymbiosis, symbiosis in which one symbiont lives inside another, is woven throughout the history of life and the story of its evolution. From the mitochondrion residing in almost every eukaryotic cell to the gut microbiome found in every human, endosymbiosis is a cornerstone of the biological processes that sustain life on Earth. While endosym-biosis is ubiquitous, many questions about its origins remain shrouded in mystery; one question in particular regards the general conditions and possible trajectories for its evolution. Modern science has hypothesized two possible pathways for the evolution of mutualistic endosymbiosis: one where an obligate antagonism is co-opted into an obligate mutualism (Co-Opted Antagonism Hypothesis), and one where a facultative mutualism evolves into an obligate mutualism (Black Queen Hypothesis). We investigated the viability of these pathways under different environmental conditions by expanding on the evolutionary agent-based system Symbulation. Specifically, we considered the impact of ectosymbiosis on de novo evolution of obligate mutualistic endosymbiosis. We found that introducing a facultative ectosymbiotic state allows endosym-biosis to evolve in a more diverse set of environmental conditions, while also decreasing the evolution of endosymbiosis in conditions where it can evolve independently.

String is a new computer language designed specifically for the implementation of ‘ribozymes’, the active entities within a new (highly simplified) model of protocellular life. The purpose of the model (which is presented here, only in outline) is the study of the abstract nature of simple cellular life and its relationship to computation. This model contains passive and active entities; passive entities are data and active ones are executable data (or programs). All programs in our model are written or evolved in String. In this paper, we describe String and provide examples of both hand-written and evolved String programs belonging to different functional categories needed for cellular operation (e.g., mass transporter, information transporter, transformer, replicator and translator). Results from the evolutionary runs are presented and discussed, where almost all ribozymes reached their optimum fitness.

Many models of organism navigation concern themselves in essence just with the sequence of locations visited and how to manage it. However, larger and bulkier organisms have also to deal with managing momentum. We expect that this affects the cognitive management of movement. Here we propose a simple model for the information processing complexity of navigation when velocity and acceleration are considered, moving away from a kinematic perspective to a partially dynamic model, to separate the effects of location and momentum management. The work is discussed in the context of recent neurobiological research suggesting that biological agents plan around acceleration and deceleration phases, showing high neural activity during their body’s velocity changes.

In this extended abstract two novel concepts are defined in the study of Random Boolean Networks, i.e. those of “pseudoattractors” and “common sea”, and it is shown how their analogues can be measured in experimental data on gene expression in single cells.

Bacteriophages, also known as phages, are viruses that infect bacteria. They are found everywhere in nature, playing vital roles in microbiomes and bacterial evolution due to the selective pressure that they place on their hosts. As obligate endosymbionts, phages depend on bacteria for successful reproduction, and either destroy their hosts through lysis or are maintained within the host through lysogeny. Lysis involves reproduction within the host cell and ultimately results in the disruption or bursting of the cell to release phage progeny. Alternatively, lysogeny is the process by which phage DNA is incorporated into the host DNA or maintained alongside the host chromosome, and thus the phage reproduces when their host reproduces. Recent work has demonstrated that phages can exist along the parasitism-mutualism spectrum, prompting questions of how phage would evolve one reproductive strategy over the other, and in which conditions. In this work, we present an agent-based model of bacteriophage/bacterial co-evolution that enables lysogenized phage to directly impact their host’s fitness by using the software platform Sym-bulation. We demonstrate that a viral population with beneficial lysogenic phage can select against lytic strategies. This result has implications for bottom-up control of vital ecosystems.

The success of deep learning is to some degree based on our ability to train models quickly using GPU or TPU hardware accelerators. Markov Brains, which are also a form of neural networks, could benefit from such an acceleration as well. However, Markov Brains are optimized using genetic algorithms, which present an even higher demand on the acceleration hardware: Not only inputs to the network and its outputs need to be communicated but new network configurations have to be loaded and tested repeatedly in large numbers. FPGAs are a natural substrate to implement Markov Brains, who are already made from deterministic logic gates. Here a Markov Brain hardware accelerator is implemented and tested, showing that Markov Brains can be computed within a single clock cycle, the ultimate hardware acceleration. However, how current FPGA design and supporting development toolchains are limiting factors, and if there is a future size speed trade-off are explored here as well.

Congestion control algorithms are used to help prevent congestion from occurring on the Internet. However, a definitive congestion control algorithm has yet to be developed. There are three reasons for this: First, the environment and usage of the Internet continue to evolve over time. Second, it is not clear what congestion control algorithms will be required as the environment evolves. Third, there is a limit to the number of the congestion control algorithms that can be developed by researchers. This paper proposes a method for automatically generating diverse congestion control algorithms and optimizing them in various environments by co-evolving network simulations as environments and congestion control algorithms as agents. In experiments conducted using co-evolution, although the algorithms generated were not on par with conventional practical congestion control algorithms, the intent of the procedures in the algorithms was interpretable from a human perspective. Furthermore, our results verify that it is possible to automatically discover a suitable environment for the evolution of a congestion control algorithm.

Genome-wide association studies (GWAS) are a powerful tool for identifying genes. They exploit the standing genetic variation and correlate phenotypic diversity to genetic markers close to or with genes of interest. However, their power is limited when it comes to complex phenotypes caused by highly epistatically interacting genes. To improve GWAS and to develop new methods, a computational model system could prove invaluable. In the computational model system presented here, the functionality of all genes in question can be identified using knockouts. This allows the comparison between the quantitative genetics results and the functional analysis. Here the goal is to perform a pilot study to investigate to which degree such a computational model can serve as a positive control for a GWAS. Surprisingly, even though the model used here is relatively simple and uses only a few genes, the GWAS struggles to identify all relevant genes. The advantages and limitations of this approach will be discussed to improve the model for future comparisons.

With the introduction of Artificial Intelligence (AI) and related technologies in our daily lives, fear and anxiety about their misuse, as well as the hidden biases in their creation, have led to a demand for regulation to address such issues. Yet, blindly regulating an innovation process that is not well understood may stifle this process and reduce benefits that society might gain from the generated technology, even under the best of intentions. Starting from a baseline game-theoretical model that captures the complex ecology of choices associated with a race for domain supremacy using AI technology, we show that socially unwanted outcomes may be produced when sanctioning is applied unconditionally to risk-taking, i.e., potentially unsafe behaviours. As an alternative to resolve the detrimental effect of over-regulation, we propose a voluntary commitment approach, wherein technologists have the freedom of choice between independently pursuing their course of actions or else establishing binding agreements to act safely, with sanctioning of those that do not abide to what they have pledged. Overall, our work reveals for the first time how voluntary commitments, with sanctions either by peers or by an institution, leads to socially beneficial outcomes in all scenarios that can be envisaged in the short-term race towards domain supremacy through AI technology.

We hypothesize that the emergence of consciousness in humans is directly related to the complexity, number of, and evolution of specialized cognitive systems. Here, we present our rationale and plan for an ongoing project to investigate the pathway to the emergence of consciousness via computer simulations of humans’ evolutionary niche using artificial-intelligence agents. Agents will contain subsets of the specialized cognitive systems and will complete tasks modeled after pressures encountered by early humans. We will observe whether the increase in cognitive complexity, measured by the number and complexity of the specialized cognitive systems, leads to an increase in task performance.

Before embarking on a new collective venture, it is important to understand partners’ preferences and intentions and how strongly they commit to a common goal. Arranging prior commitments of future actions has been shown to be an evolutionary viable strategy in the context of social dilemmas. Previous works have focused on simple well-mixed population settings, for ease of analysis. Here, starting from a baseline model of a coordination game with asymmetric benefits for technology adoption in the well-mixed setting, we examine the impact of different population structures, including square lattice and scale-free (SF) networks, capturing typical homogeneous and heterogeneous network structures, on the dynamics of decision-making in the context of coordinating technology adoption. We show that, similarly to previous well-mixed analyses, prior commitments enhance coordination and the overall population payoff in structured populations, especially when the cost of commitment is justified against the benefit of coordination, and when the technology market is highly competitive. When commitments are absent, slightly higher levels of coordination and population welfare are obtained in SF than lattice. In the presence of commitments and when the market is very competitive, the overall population welfare is similar in both lattice and heterogeneous networks; though it is slightly lower in SF when the market competition is low, while social welfare suffers in a monopolistic setting. Overall, we observe that commitments can improve coordination and population welfare in structured populations, but in its presence, the outcome of evolutionary dynamics is, interestingly, not sensitive to changes in the network structure.

Indirect reciprocity (IR) is an important mechanism for promoting cooperation among self-interested agents. Simplified, it means: “you help me, therefore somebody else will help you” (in contrast to direct reciprocity: “you help me; therefore I will help you”). IR can be achieved via reputation and norms. However, it was often argued that IR only works if reputations are public and does not do so under private assessment (PriA). Yet, recent papers suggest that IR under PriA is feasible, and that it has more variety and ways to improve, than have been considered before.

Our modern world is teeming with non-biological agents, whose growing complexity brings them so close to living beings that they can be cataloged as artificial creatures, i.e., a form of Artificial Life (ALife). Ranging from disembodied intelligent agents to robots of conspicuous dimensions, all these artifacts are united by the fact that they are designed, built, and possibly trained by humans taking inspiration from natural elements. Hence, humans play a fundamental role in relation to ALife, both as creators and as final users, which calls attention to the need of studying the mutual influence of human and artificial life. Here we attempt an experimental investigation of the reciprocal effects of the human-ALife interaction. To this extent, we design an artificial world populated by life-like creatures, and resort to open-ended evolution to foster the creatures adaptation. We allow bidirectional communication between the system and humans, who can observe the artificial world and voluntarily choose to perform positive or negative actions towards the creatures populating it; those actions may have a short- or long-term impact on the artificial creatures. Our experimental results show that the creatures are capable of evolving under the influence of humans, even though the impact of the interaction remains uncertain. In addition, we find that ALife gives rise to disparate feelings in humans who interact with it, who are not always aware of the importance of their conduct.