Male and female contributions to diversity among birdwing butterfly images.

Machine learning (ML) newly enables tests for higher inter-species diversity in visible phenotype (disparity) among males versus females, predictions made from Darwinian sexual selection versus Wallacean natural selection, respectively. Here, we use ML to quantify variation across a sample of > 16,000 dorsal and ventral photographs of the sexually dimorphic birdwing butterflies (Lepidoptera: Papilionidae). Validation of image embedding distances, learnt by a triplet-trained, deep convolutional neural network, shows ML can be used for automated reconstruction of phenotypic evolution achieving measures of phylogenetic congruence to genetic species trees within a range sampled among genetic trees themselves. Quantification of sexual disparity difference (male versus female embedding distance), shows sexually and phylogenetically variable inter-species disparity. Ornithoptera exemplify high embedded male image disparity, diversification of selective optima in fitted multi-peak OU models and accelerated divergence, with cases of extreme divergence in allopatry and sympatry. However, genus Troides shows inverted patterns, including comparatively static male embedded phenotype, and higher female than male disparity - though within an inferred selective regime common to these females. Birdwing shapes and colour patterns that are most phenotypically distinctive in ML similarity are generally those of males. However, either sex can contribute majoritively to observed phenotypic diversity among species.

Erratum: Guttenberg et al. Classification of the Biogenicity of Complex Organic Mixtures for the Detection of Extraterrestrial Life. Life 2021, 11, 234.

There was an error in the original article [...].

Classification of the Biogenicity of Complex Organic Mixtures for the Detection of Extraterrestrial Life.

Searching for life in the Universe depends on unambiguously distinguishing biological features from background signals, which could take the form of chemical, morphological, or spectral signatures. The discovery and direct measurement of organic compounds unambiguously indicative of extraterrestrial (ET) life is a major goal of Solar System exploration. Biology processes matter and energy differently from abiological systems, and materials produced by biological systems may become enriched in planetary environments where biology is operative. However, ET biology might be composed of different components than terrestrial life. As ET sample return is difficult, in situ methods for identifying biology will be useful. Mass spectrometry (MS) is a potentially versatile life detection technique, which will be used to analyze numerous Solar System environments in the near future. We show here that simple algorithmic analysis of MS data from abiotic synthesis (natural and synthetic), microbial cells, and thermally processed biological materials (lab-grown organisms and petroleum) easily identifies relational organic compound distributions that distinguish pristine and aged biological and abiological materials, which likely can be attributed to the types of compounds these processes produce, as well as how they are formed and decompose. To our knowledge this is the first comprehensive demonstration of the utility of this analytical technique for the detection of biology. This method is independent of the detection of particular masses or molecular species samples may contain. This suggests a general method to agnostically detect evidence of biology using MS given a sufficiently strong signal in which the majority of the material in a sample has either a biological or abiological origin. Such metrics are also likely to be useful for studies of possible emergent living phenomena, and paleobiological samples.

Cross-Subject EEG-Based Emotion Recognition Through Neural Networks With Stratified Normalization.

Due to a large number of potential applications, a good deal of effort has been recently made toward creating machine learning models that can recognize evoked emotions from one's physiological recordings. In particular, researchers are investigating the use of EEG as a low-cost, non-invasive method. However, the poor homogeneity of the EEG activity across participants hinders the implementation of such a system by a time-consuming calibration stage. In this study, we introduce a new participant-based feature normalization method, named stratified normalization, for training deep neural networks in the task of cross-subject emotion classification from EEG signals. The new method is able to subtract inter-participant variability while maintaining the emotion information in the data. We carried out our analysis on the SEED dataset, which contains 62-channel EEG recordings collected from 15 participants watching film clips. Results demonstrate that networks trained with stratified normalization significantly outperformed standard training with batch normalization. In addition, the highest model performance was achieved when extracting EEG features with the multitaper method, reaching a classification accuracy of 91.6% for two emotion categories (positive and negative) and 79.6% for three (also neutral). This analysis provides us with great insight into the potential benefits that stratified normalization can have when developing any cross-subject model based on EEG.

Impacts of speciation and extinction measured by an evolutionary decay clock.

The hypothesis that destructive mass extinctions enable creative evolutionary radiations (creative destruction) is central to classic concepts of macroevolution1,2. However, the relative impacts of extinction and radiation on the co-occurrence of species have not been directly quantitatively compared across the Phanerozoic eon. Here we apply machine learning to generate a spatial embedding (multidimensional ordination) of the temporal co-occurrence structure of the Phanerozoic fossil record, covering 1,273,254 occurrences in the Paleobiology Database for 171,231 embedded species. This facilitates the simultaneous comparison of macroevolutionary disruptions, using measures independent of secular diversity trends. Among the 5% most significant periods of disruption, we identify the 'big five' mass extinction events2, seven additional mass extinctions, two combined mass extinction-radiation events and 15 mass radiations. In contrast to narratives that emphasize post-extinction radiations1,3, we find that the proportionally most comparable mass radiations and extinctions (such as the Cambrian explosion and the end-Permian mass extinction) are typically decoupled in time, refuting any direct causal relationship between them. Moreover, in addition to extinctions4, evolutionary radiations themselves cause evolutionary decay (modelled co-occurrence probability and shared fraction of species between times approaching zero), a concept that we describe as destructive creation. A direct test of the time to over-threshold macroevolutionary decay4 (shared fraction of species between two times ≤ 0.1), counted by the decay clock, reveals saw-toothed fluctuations around a Phanerozoic mean of 18.6 million years. As the Quaternary period began at a below-average decay-clock time of 11 million years, modern extinctions further increase life's decay-clock debt.

Information generation as a functional basis of consciousness.

What is the biological advantage of having consciousness? Functions of consciousness have been elusive due to the subjective nature of consciousness and ample empirical evidence showing the presence of many nonconscious cognitive performances in the human brain. Drawing upon empirical literature, here, we propose that a core function of consciousness be the ability to internally generate representations of events possibly detached from the current sensory input. Such representations are constructed by generative models learned through sensory-motor interactions with the environment. We argue that the ability to generate information underlies a variety of cognitive functions associated with consciousness such as intention, imagination, planning, short-term memory, attention, curiosity, and creativity, all of which contribute to non-reflexive behavior. According to this view, consciousness emerged in evolution when organisms gained the ability to perform internal simulations using internal models, which endowed them with flexible intelligent behavior. To illustrate the notion of information generation, we take variational autoencoders (VAEs) as an analogy and show that information generation corresponds the decoding (or decompression) part of VAEs. In biological brains, we propose that information generation corresponds to top-down predictions in the predictive coding framework. This is compatible with empirical observations that recurrent feedback activations are linked with consciousness whereas feedforward processing alone seems to occur without evoking conscious experience. Taken together, the information generation hypothesis captures many aspects of existing ideas about potential functions of consciousness and provides new perspectives on the functional roles of consciousness.

Deep learning on butterfly phenotypes tests evolution's oldest mathematical model.

Traditional anatomical analyses captured only a fraction of real phenomic information. Here, we apply deep learning to quantify total phenotypic similarity across 2468 butterfly photographs, covering 38 subspecies from the polymorphic mimicry complex of Heliconius erato and Heliconius melpomene. Euclidean phenotypic distances, calculated using a deep convolutional triplet network, demonstrate significant convergence between interspecies co-mimics. This quantitatively validates a key prediction of Müllerian mimicry theory, evolutionary biology's oldest mathematical model. Phenotypic neighbor-joining trees are significantly correlated with wing pattern gene phylogenies, demonstrating objective, phylogenetically informative phenome capture. Comparative analyses indicate frequency-dependent mutual convergence with coevolutionary exchange of wing pattern features. Therefore, phenotypic analysis supports reciprocal coevolution, predicted by classical mimicry theory but since disputed, and reveals mutual convergence as an intrinsic generator for the unexpected diversity of Müllerian mimicry. This demonstrates that deep learning can generate phenomic spatial embeddings, which enable quantitative tests of evolutionary hypotheses previously only testable subjectively.

On the Potential for Open-Endedness in Neural Networks.

Natural evolution gives the impression of leading to an open-ended process of increasing diversity and complexity. If our goal is to produce such open-endedness artificially, this suggests an approach driven by evolutionary metaphor. On the other hand, techniques from machine learning and artificial intelligence are often considered too narrow to provide the sort of exploratory dynamics associated with evolution. In this article, we hope to bridge that gap by reviewing common barriers to open-endedness in the evolution-inspired approach and how they are dealt with in the evolutionary case-collapse of diversity, saturation of complexity, and failure to form new kinds of individuality. We then show how these problems map onto similar ones in the machine learning approach, and discuss how the same insights and solutions that alleviated those barriers in evolutionary approaches can be ported over. At the same time, the form these issues take in the machine learning formulation suggests new ways to analyze and resolve barriers to open-endedness. Ultimately, we hope to inspire researchers to be able to interchangeably use evolutionary and gradient-descent-based machine learning methods to approach the design and creation of open-ended systems.

Swarming in the Dirt: Ordered Flocks with Quenched Disorder.

The effect of quenched (frozen) disorder on the collective motion of active particles is analyzed. We find that active polar systems are far more robust against quenched disorder than equilibrium ferromagnets. Long-ranged order (a nonzero average velocity ⟨v⟩) persists in the presence of quenched disorder even in spatial dimensions d=3; in d=2, quasi-long-ranged order (i.e., spatial velocity correlations that decay as a power law with distance) occurs. In equilibrium systems, only quasi-long-ranged order in d=3 and short-ranged order in d=2 are possible. Our theoretical predictions for two dimensions are borne out by simulations.

Bulk measurements of messy chemistries are needed for a theory of the origins of life.

A feature of many of the chemical systems plausibly involved in the origins of terrestrial life is that they are complex and messy-producing a wide range of compounds via a wide range of mechanisms. However, the fundamental behaviour of such systems is currently not well understood; we do not have the tools to make statistical predictions about such complex chemical networks. This is, in part, due to a lack of quantitative data from which such a theory could be built; specifically, functional measurements of messy chemical systems. Here, we propose that the pantheon of experimental approaches to the origins of life should be expanded to include the study of 'functional measurements'-the direct study of bulk properties of chemical systems and their interactions with other compounds, the formation of structures and other behaviours, even in cases where the precise composition and mechanisms are unknown.This article is part of the themed issue 'Reconceptualizing the origins of life'.

Transferable Measurements of Heredity in Models of the Origins of Life.

We propose a metric which can be used to compute the amount of heritable variation enabled by a given dynamical system. A distribution of selection pressures is used such that each pressure selects a particular fixed point via competitive exclusion in order to determine the corresponding distribution of potential fixed points in the population dynamics. This metric accurately detects the number of species present in artificially prepared test systems, and furthermore can correctly determine the number of heritable sets in clustered transition matrix models in which there are no clearly defined genomes. Finally, we apply our metric to the GARD model and show that it accurately reproduces prior measurements of the model's heritability.

Macroscopic traveling packet and soliton states of quasi-one-dimensional flocks.

Using a continuum model for inhomogeneous flocks, we show that a finite but arbitrarily large moving "packet" of active particles (e.g., moving creatures) can form in a background of a lower density disordered phase of these particles, like a liquid drop surrounded by vapor. The "vapor density" of the disordered background can be made arbitrarily low. We find three basic types of quasi-one-dimensional states: "longitudinal", "transverse", and "oblique" states, with their internal velocity fields, respectively, parallel, perpendicular, and oblique to the interface. The transitions between these states are also studied.

Still water: dead zones and collimated ejecta from the impact of granular jets.

When a dense granular jet hits a target, it forms a large dead zone and ejects a highly collimated conical sheet with a well-defined opening angle. Using experiments, simulations, and continuum modeling, we find that this opening angle is insensitive to the precise target shape and the dissipation mechanisms in the flow. We show that this surprising insensitivity arises because dense granular jet impact, though highly dissipative, is nonetheless controlled by the limit of perfect fluid flow.

Minimizing memory as an objective for coarse-graining.

Coarse-graining a molecular model is the process of integrating over degrees of freedom to obtain a reduced representation. This process typically involves two separate but related steps, selection of the coordinates comprising the reduced system and modeling their interactions. Both the coordinate selection and the modeling procedure present challenges. Here, we focus on the former. Typically, one seeks to integrate over the fast degrees of freedom and retain the slow degrees of freedom. Failure to separate timescales results in memory. With this motivation, we introduce a heuristic measure of memory and show that it can be used to compare competing coordinate selections for a given modeling procedure. We numerically explore the utility of this heuristic for three systems of increasing complexity. The first example is a four-particle linear model, which is exactly solvable. The second example is a sixteen-particle nonlinear model; this system has interactions that are characteristic of molecular force fields but is still sufficiently simple to permit exhaustive numerical treatment. The third example is an atomic-resolution representation of a protein, the class of models most often treated by relevant coarse-graining approaches; we specifically study an actin monomer. In all three cases, we find that the heuristic suggests coordinate selections that are physically intuitive and reflect molecular structure. The memory heuristic can thus serve as an objective codification of expert knowledge and a guide to sites within a model that requires further attention.

Microscopic dissipation in a cohesionless granular jet impact.

Sufficiently fine granular systems appear to exhibit continuum properties, though the precise continuum limit obtained can be vastly different depending on the particular system. In the present paper the continuum limit of an unconfined, dense granular flow is investigated. To do this a two-dimensional dense cohesionless granular jet impinging upon a target is used as a test system. This is simulated via a time-step-driven hard-sphere method and apply a mean-field theoretical approach to connect the macroscopic flow with the microscopic material parameters of the grains. It is observed that the flow separates into a cone with an interior cone angle determined by the conservation of momentum and the dissipation of energy. From the cone angle a dimensionless quantity A-B that characterizes the flow is extracted. This quantity is found to depend both on whether or not a dead zone, i.e., a stationary region near the target, is present and on the value of the coefficient of dynamic friction. A theory is presented for the scaling of A-B with the coefficient of friction that suggests that dissipation is primarily a perturbative effect in this flow rather than the source of qualitatively different behavior.

Steered transition path sampling.

We introduce a path sampling method for obtaining statistical properties of an arbitrary stochastic dynamics. The method works by decomposing a trajectory in time, estimating the probability of satisfying a progress constraint, modifying the dynamics based on that probability, and then reweighting to calculate averages. Because the progress constraint can be formulated in terms of occurrences of events within time intervals, the method is particularly well suited for controlling the sampling of currents of dynamic events. We demonstrate the method for calculating transition probabilities in barrier crossing problems and survival probabilities in strongly diffusive systems with absorbing states, which are difficult to treat by shooting. We discuss the relation of the algorithm to other methods.

Short-time evolution in the adaptive immune system.

We exploit a simple model to numerically and analytically investigate the effect of enforcing a time constraint for achieving a system-wide goal during an evolutionary dynamics. This situation is relevant to finding antibody specificities in the adaptive immune response as well as to artificial situations in which an evolutionary dynamics is used to generate a desired capability in a limited number of generations. When the likelihood of finding the target phenotype is low, we find that the optimal mutation rate can exceed the error threshold, in contrast to conventional evolutionary dynamics. We also show how a logarithmic correction to the usual inverse scaling of population size with mutation rate arises. Implications for natural and artificial evolutionary situations are discussed.

Approximate hard-sphere method for densely packed granular flows.

The simulation of granular media is usually done either with event-driven codes that treat collisions as instantaneous but have difficulty with very dense packings, or with molecular dynamics (MD) methods that approximate rigid grains using a stiff viscoelastic spring. There is a little-known method that combines several collision events into a single timestep to retain the instantaneous collisions of event-driven dynamics, but also be able to handle dense packings. However, it is poorly characterized as to its regime of validity and failure modes. We present a modification of this method to reduce the introduction of overlap error, and test it using the problem of two-dimensional (2D) granular Couette flow, a densely packed system that has been well characterized by previous work. We find that this method can successfully replicate the results of previous work up to the point of jamming, and that it can do so a factor of 10 faster than comparable MD methods.

Dynamics of gene duplication and transposons in microbial genomes following a sudden environmental change.

The impact of gene duplication on evolution is as ubiquitous as point mutation, but this realization is not yet reflected in our quantitative models of population genetics. In this Commentary article, we explore the implications of such models of gene duplication, specifically expanding on our previous work. We lay down a framework for understanding the impact of gene duplications on the evolution a biological network and give an analytical argument based on the concept of mutational error threshold for the necessity of gene duplications for the evolution of complex networks. In other words, by realizing that the impact of mutations must act appropriately in order to allow for the maintenance of complex networks, we develop a mathematical scaling argument that shows why gene duplication provides the types of mutations more favorable to increasing complexity. In the process of doing so, we seek to explain the relationship between per base pair mutation rates and genome size.