Recording animal-view videos of the natural world using a novel camera system and software package.

Plants, animals, and fungi display a rich tapestry of colors. Animals, in particular, use colors in dynamic displays performed in spatially complex environments. Although current approaches for studying colors are objective and repeatable, they miss the temporal variation of color signals entirely. Here, we introduce hardware and software that provide ecologists and filmmakers the ability to accurately record animal-perceived colors in motion. Specifically, our Python codes transform photos or videos into perceivable units (quantum catches) for animals of known photoreceptor sensitivity. The plans and codes necessary for end-users to capture animal-view videos are all open source and publicly available to encourage continual community development. The camera system and the associated software package will allow ecologists to investigate how animals use colors in dynamic behavioral displays, the ways natural illumination alters perceived colors, and other questions that remained unaddressed until now due to a lack of suitable tools. Finally, it provides scientists and filmmakers with a new, empirically grounded approach for depicting the perceptual worlds of nonhuman animals.

Through an animal's eye: the implications of diverse sensory systems in scientific experimentation.

'Accounting for the sensory abilities of animals is critical in experimental design.' No researcher would disagree with this statement, yet it is often the case that we inadvertently fall for anthropocentric biases and use ourselves as the reference point. This paper discusses the risks of adopting an anthropocentric view when working with non-human animals, and the unintended consequences this has on our experimental designs and results. To this aim, we provide general examples of anthropocentric bias from different fields of animal research, with a particular focus on animal cognition and behaviour, and lay out the potential consequences of adopting a human-based perspective. Knowledge of the sensory abilities, both in terms of similarities to humans and peculiarities of the investigated species, is crucial to ensure solid conclusions. A more careful consideration of the diverse sensory systems of animals would improve many scientific fields and enhance animal welfare in the laboratory.

Assessing preferences for adult versus juvenile features in young animals: Newly hatched chicks spontaneously approach red and large stimuli.

Young precocial birds benefit from staying close to both their mother and their siblings, while prioritising adults, which provide better care. Which features of the stimuli are used by young birds to prioritise approach and eventually attachment to adults over siblings is unknown. We started to address this question in newly hatched domestic chicks (Gallus gallus), focusing on their spontaneous preferences for visual features that systematically vary between adult and juvenile chickens, and that had previously been identified as attractive: size (larger in adults than in juveniles) and colour (darker and redder in adults than in juveniles). Overall, chicks at their first visual experience, that had never seen a conspecific beforehand, were most attracted to the red and large stimuli (two adult features) and spent more time in close proximity with red stimuli than with yellow stimuli. When tested with red large versus small objects (Exp. 1), chicks preferred the large shape. When tested with yellow large and small objects (Exp. 2), chicks did not show a preference. Chicks had a stronger preference for large red stimuli (vs. small yellow objects) than for small red stimuli (vs. a large yellow object) (Exp. 3). These results suggest that the combination of size and colour form the predisposition that helps chicks to spontaneously discriminate between adult and juvenile features from the first stages of life, in the absence of previous experience, exhibiting a preference to approach stimuli with features associated with the presence of adult conspecifics.

A spontaneous gravity prior: newborn chicks prefer stimuli that move against gravity.

At the beginning of life, inexperienced animals use evolutionary-given preferences (predispositions) to decide what stimuli to attend and approach. Stimuli that contain cues of animacy, such as face-like stimuli, biological motion and changes in speed, are particularly attractive across vertebrate taxa. A strong cue of animacy is upward movement against terrestrial gravity, because only animate objects consistently move upward. To test whether upward movement is spontaneously considered attractive already at birth, we tested the early preferences of dark-hatched chicks (Gallus gallus) for upward- versus downward-moving visual stimuli. We found that, without any previous visual experience, chicks consistently exhibited a preference to approach stimuli that move upward, against gravity. A control experiment showed that these preferences are not driven by avoidance of downward stimuli. These results show that newborn animals have a gravity prior that attracts them toward upward movement. Movement against gravity can be used as a cue of animacy to orient early approach responses in the absence of previous visual experience.

Multispectral images of flowers reveal the adaptive significance of using long-wavelength-sensitive receptors for edge detection in bees.

Many pollinating insects acquire their entire nutrition from visiting flowers, and they must therefore be efficient both at detecting flowers and at recognizing familiar rewarding flower types. A crucial first step in recognition is the identification of edges and the segmentation of the visual field into areas that belong together. Honeybees and bumblebees acquire visual information through three types of photoreceptors; however, they only use a single receptor type-the one sensitive to longer wavelengths-for edge detection and movement detection. Here, we show that these long-wavelength receptors (peak sensitivity at ~544 nm, i.e., green) provide the most consistent signals in response to natural objects. Using our multispectral image database of flowering plants, we found that long-wavelength receptor responses had, depending on the specific scenario, up to four times higher signal-to-noise ratios than the short- and medium-wavelength receptors. The reliability of the long-wavelength receptors emerges from an intricate interaction between flower coloration and the bee's visual system. This finding highlights the adaptive significance of bees using only long-wavelength receptors to locate flowers among leaves, before using information provided by all three receptors to distinguish the rewarding flower species through trichromatic color vision.

Primordial evolvability: Impasses and challenges.

While it is generally agreed that some kind of replicating non-living compounds were the precursors of life, there is much debate over their possible chemical nature. Metabolism-first approaches propose that mutually catalytic sets of simple organic molecules could be capable of self-replication and rudimentary chemical evolution. In particular, the graded autocatalysis replication domain (GARD) model, depicting assemblies of amphiphilic molecules, has received considerable interest. The system propagates compositional information across generations and is suggested to be a target of natural selection. However, evolutionary simulations indicate that the system lacks selectability (i.e. selection has negligible effect on the equilibrium concentrations). We elaborate on the lessons learnt from the example of the GARD model and, more widely, on the issue of evolvability, and discuss the implications for similar metabolism-first scenarios. We found that simple incorporation-type chemistry based on non-covalent bonds, as assumed in GARD, is unlikely to result in alternative autocatalytic cycles when catalytic interactions are randomly distributed. An even more serious problem stems from the lognormal distribution of catalytic factors, causing inherent kinetic instability of such loops, due to the dominance of efficiently catalyzed components that fail to return catalytic aid. Accordingly, the dynamics of the GARD model is dominated by strongly catalytic, but not auto-catalytic, molecules. Without effective autocatalysis, stable hereditary propagation is not possible. Many repetitions and different scaling of the model come to no rescue. Despite all attempts to show the contrary, the GARD model is not evolvable, in contrast to reflexively autocatalytic networks, complemented by rare uncatalyzed reactions and compartmentation. The latter networks, resting on the creation and breakage of chemical bonds, can generate novel ('mutant') autocatalytic loops from a given set of environmentally available compounds. Real chemical reactions that make or break covalent bonds, rather than mere incorporation of components, are necessary for open-ended evolvability. The issue of whether or not several concrete chemical systems (rather than singular curiosities) could realize reflexively autocatalytic macromolecular networks will ultimately determine the relevance of metabolism-first approaches to the origin of life, as stepping stones towards true open-endedness that requires the combination of rich combinatorial chemistry controlled by information stored in template replicators.

Spontaneous biases enhance generalization in the neonate brain.

Inductive generalization is adaptive in novel contexts for both biological and artificial intelligence. Spontaneous generalization in inexperienced animals raises questions on whether predispositions (evolutionarily acquired biases, or priors) enable generalization from sparse data, without reinforcement. We exposed neonate chicks to an artificial social partner of a specific color, and then looked at generalization on the red-yellow or blue-green ranges. Generalization was inconsistent with an unbiased model. Biases included asymmetrical generalization gradients, some preferences for unfamiliar stimuli, different speed of learning, faster learning for colors infrequent in the natural spectrum. Generalization was consistent with a Bayesian model that incorporates predispositions as initial preferences and treats the learning process as an update of predispositions. Newborn chicks are evolutionarily prepared for generalization, via biases independent from experience, reinforcement, or supervision. To solve the problem of induction, biological and artificial intelligence can use biases tuned to infrequent stimuli, such as the red and blue colors.

Evolution of associative learning in chemical networks.

Organisms that can learn about their environment and modify their behaviour appropriately during their lifetime are more likely to survive and reproduce than organisms that do not. While associative learning - the ability to detect correlated features of the environment - has been studied extensively in nervous systems, where the underlying mechanisms are reasonably well understood, mechanisms within single cells that could allow associative learning have received little attention. Here, using in silico evolution of chemical networks, we show that there exists a diversity of remarkably simple and plausible chemical solutions to the associative learning problem, the simplest of which uses only one core chemical reaction. We then asked to what extent a linear combination of chemical concentrations in the network could approximate the ideal Bayesian posterior of an environment given the stimulus history so far? This Bayesian analysis revealed the 'memory traces' of the chemical network. The implication of this paper is that there is little reason to believe that a lack of suitable phenotypic variation would prevent associative learning from evolving in cell signalling, metabolic, gene regulatory, or a mixture of these networks in cells.

Lack of evolvability in self-sustaining autocatalytic networks constraints metabolism-first scenarios for the origin of life.

A basic property of life is its capacity to experience Darwinian evolution. The replicator concept is at the core of genetics-first theories of the origin of life, which suggest that self-replicating oligonucleotides or their similar ancestors may have been the first "living" systems and may have led to the evolution of an RNA world. But problems with the nonenzymatic synthesis of biopolymers and the origin of template replication have spurred the alternative metabolism-first scenario, where self-reproducing and evolving proto-metabolic networks are assumed to have predated self-replicating genes. Recent theoretical work shows that "compositional genomes" (i.e., the counts of different molecular species in an assembly) are able to propagate compositional information and can provide a setup on which natural selection acts. Accordingly, if we stick to the notion of replicator as an entity that passes on its structure largely intact in successive replications, those macromolecular aggregates could be dubbed "ensemble replicators" (composomes) and quite different from the more familiar genes and memes. In sharp contrast with template-dependent replication dynamics, we demonstrate here that replication of compositional information is so inaccurate that fitter compositional genomes cannot be maintained by selection and, therefore, the system lacks evolvability (i.e., it cannot substantially depart from the asymptotic steady-state solution already built-in in the dynamical equations). We conclude that this fundamental limitation of ensemble replicators cautions against metabolism-first theories of the origin of life, although ancient metabolic systems could have provided a stable habitat within which polymer replicators later evolved.

The robustness of keystone indices in food webs.

Species that have outstanding importance in the functioning of a community are called keystone species. Network indices are increasingly used to identify them, e.g. for conservation biological purposes. The problem is that the calculation of these indices is based on the particular network model of the studied food web, which can include network construction errors. For example, additional, unnecessary trophic links can be built in, or, to the contrary, functional links can be left out. What is the effect of such errors on the result of network analysis, e.g. the centrality values of species? Can you rely on the importance rank of species that you calculated? We developed a robustness measure (R) for network indices to answer these questions. R is proportional to the likeliness that the importance rank of nodes in the given network according to a given index would not change due to possible errors in network construction. For calculating R, first the maximum expected error (P) has to be computed which represents the potential range of error in estimating the keystone index in question. Basically, R is calculated by comparing P to the keystone indices of species to assess the reliability of the importance rank of species based on the network model. We calculated the robustness of 13 different structural indices in 26 food webs of different size to test the P and R values. We found that fragmentation indices and the number of dominated nodes can be characterized by quite low R values, while betweenness, topological importance, keystoneness and mixed trophic impact have high R values, which means that they are relatively more reliable for assessing the importance rank of species in an uncertain network model. However, as R was found to be very variable, depending on the topology of a given network, a detailed description is provided for performing the actual calculations case-by-case.

Insect-Inspired Sequential Inspection Strategy Enables an Artificial Network of Four Neurons to Estimate Numerosity.

Varying levels of numerical cognition have been found in several animal species. Bees, in particular, have been argued to be able to count up to four items and solve complex numerical tasks. Here we present an exceedingly simple neural circuit that, when provided with the actual visual input that the bee is receiving while carrying out the task, can make reliable estimates on the number of items in the display. Thus we suggest that the elegance of numerical problem solving in bees might not lie in the formation of numerical concepts (such as "more," "less," or "zero"), but in the use of specific flight movements to scan targets, which streamlines the visual input and so renders the task of counting computationally inexpensive. Careful examination of the actual inspection strategies used by animals might reveal that animals often employ active scanning behaviors as shortcuts to simplify complex visual pattern discrimination tasks.

Randomly weighted receptor inputs can explain the large diversity of colour-coding neurons in the bee visual system.

True colour vision requires comparing the responses of different spectral classes of photoreceptors. In insects, there is a wealth of data available on the physiology of photoreceptors and on colour-dependent behaviour, but less is known about the neural mechanisms that link the two. The available information in bees indicates a diversity of colour opponent neurons in the visual optic ganglia that significantly exceeds that known in humans and other primates. Here, we present a simple mathematical model for colour processing in the optic lobes of bees to explore how this diversity might arise. We found that the model can reproduce the physiological spectral tuning curves of the 22 neurons that have been described so far. Moreover, the distribution of the presynaptic weights in the model suggests that colour-coding neurons are likely to be wired up to the receptor inputs randomly. The perceptual distances in our random synaptic weight model are in agreement with behavioural observations. Our results support the idea that the insect nervous system might adopt partially random wiring of neurons for colour processing.

Evolution before genes.

BACKGROUND: Our current understanding of evolution is so tightly linked to template-dependent replication of DNA and RNA molecules that the old idea from Oparin of a self-reproducing 'garbage bag' ('coacervate') of chemicals that predated fully-fledged cell-like entities seems to be farfetched to most scientists today. However, this is exactly the kind of scheme we propose for how Darwinian evolution could have occurred prior to template replication. RESULTS: We cannot confirm previous claims that autocatalytic sets of organic polymer molecules could undergo evolution in any interesting sense by themselves. While we and others have previously imagined inhibition would result in selectability, we found that it produced multiple attractors in an autocatalytic set that cannot be selected for. Instead, we discovered that if general conditions are satisfied, the accumulation of adaptations in chemical reaction networks can occur. These conditions are the existence of rare reactions producing viable cores (analogous to a genotype), that sustains a molecular periphery (analogous to a phenotype). CONCLUSIONS: We conclude that only when a chemical reaction network consists of many such viable cores, can it be evolvable. When many cores are enclosed in a compartment there is competition between cores within the same compartment, and when there are many compartments, there is between-compartment competition due to the phenotypic effects of cores and their periphery at the compartment level. Acquisition of cores by rare chemical events, and loss of cores at division, allows macromutation, limited heredity and selectability, thus explaining how a poor man's natural selection could have operated prior to genetic templates. This is the only demonstration to date of a mechanism by which pre-template accumulation of adaptation could occur.

Evolvable neuronal paths: a novel basis for information and search in the brain.

We propose a previously unrecognized kind of informational entity in the brain that is capable of acting as the basis for unlimited hereditary variation in neuronal networks. This unit is a path of activity through a network of neurons, analogous to a path taken through a hidden Markov model. To prove in principle the capabilities of this new kind of informational substrate, we show how a population of paths can be used as the hereditary material for a neuronally implemented genetic algorithm, (the swiss-army knife of black-box optimization techniques) which we have proposed elsewhere could operate at somatic timescales in the brain. We compare this to the same genetic algorithm that uses a standard 'genetic' informational substrate, i.e. non-overlapping discrete genotypes, on a range of optimization problems. A path evolution algorithm (PEA) is defined as any algorithm that implements natural selection of paths in a network substrate. A PEA is a previously unrecognized type of natural selection that is well suited for implementation by biological neuronal networks with structural plasticity. The important similarities and differences between a standard genetic algorithm and a PEA are considered. Whilst most experiments are conducted on an abstract network model, at the conclusion of the paper a slightly more realistic neuronal implementation of a PEA is outlined based on Izhikevich spiking neurons. Finally, experimental predictions are made for the identification of such informational paths in the brain.