Empirical research has shown that when making choices based on probabilistic options, people behave as if they overestimate small probabilities, underestimate large probabilities, and treat positive and negative outcomes differently. These distortions have been modeled using a nonlinear probability weighting function, which is found in several nonexpected utility theories, including rank-dependent models and prospect theory; here, we propose a Bayesian approach to the probability weighting function and, with it, a psychological rationale. In the real world, uncertainty is ubiquitous and, accordingly, the optimal strategy is to combine probability statements with prior information using Bayes' rule. First, we show that any reasonable prior on probabilities leads to 2 of the observed effects; overweighting of low probabilities and underweighting of high probabilities. We then investigate 2 plausible kinds of priors: informative priors based on previous experience and uninformative priors of ignorance. Individually, these priors potentially lead to large problems of bias and inefficiency, respectively; however, when combined using Bayesian model comparison methods, both forms of prior can be applied adaptively, gaining the efficiency of empirical priors and the robustness of ignorance priors. We illustrate this for the simple case of generic good and bad options, using Internet blogs to estimate the relevant priors of inference. Given this combined ignorant/informative prior, the Bayesian probability weighting function is not only robust and efficient but also matches all of the major characteristics of the distortions found in empirical research.

Avoiding detection can provide significant survival advantages for prey, predators, or the military; conversely, maximizing visibility would be useful for signalling. One simple determinant of detectability is an animal's colour relative to its environment. But identifying the optimal colour to minimize (or maximize) detectability in a given natural environment is complex, partly because of the nature of the perceptual space. Here for the first time, using image processing techniques to embed targets into realistic environments together with psychophysics to estimate detectability and deep neural networks to interpolate between sampled colours, we propose a method to identify the optimal colour that either minimizes or maximizes visibility. We apply our approach in two natural environments (temperate forest and semi-arid desert) and show how a comparatively small number of samples can be used to predict robustly the most and least effective colours for camouflage. To illustrate how our approach can be generalized to other non-human visual systems, we also identify the optimum colours for concealment and visibility when viewed by simulated red–green colour-blind dichromats, typical for non-human mammals. Contrasting the results from these visual systems sheds light on why some predators seem, at least to humans, to have colouring that would appear detrimental to ambush hunting. We found that for simulated dichromatic observers, colour strongly affected detection time for both environments. In contrast, trichromatic observers were more effective at breaking camouflage.

Evolutionary biologists frequently wish to measure the fitness of alternative phenotypes using behavioral experiments. However, many phenotypes are complex. One example is coloration: camouflage aims to make detection harder, while conspicuous signals (e.g., for warning or mate attraction) require the opposite. Identifying the hardest and easiest to find patterns is essential for understanding the evolutionary forces that shape protective coloration, but the parameter space of potential patterns (colored visual textures) is vast, limiting previous empirical studies to a narrow range of phenotypes. Here, we demonstrate how deep learning combined with genetic algorithms can be used to augment behavioral experiments, identifying both the best camouflage and the most conspicuous signal(s) from an arbitrarily vast array of patterns. To show the generality of our approach, we do so for both trichromatic (e.g., human) and dichromatic (e.g., typical mammalian) visual systems, in two different habitats. The patterns identified were validated using human participants; those identified as the best for camouflage were significantly harder to find than a tried-and-tested military design, while those identified as most conspicuous were significantly easier to find than other patterns. More generally, our method, dubbed the "Camouflage Machine," will be a useful tool for identifying the optimal phenotype in high dimensional state spaces.

1. One of the most challenging issues in modelling the evolution of protective colouration is the immense number of potential combinations of colours and textures.2. We describe CamoGAN, a novel method to exploit Generative Adversarial Networks to simulate an evolutionary arms race between the camouflage of a synthetic prey and its predator.3. Patterns evolved using our methods are shown to provide progressively more effective concealment and outperform two recognized camouflage techniques, as validated by using humans as visual predators. 4. We believe CamoGAN will be highly useful, particularly for biologists, for rapidly developing and testing optimal camouflage or signalling patterns in multiple environments. K E Y W O R D SCamoGAN, camouflage, co-evolution, deep learning, generative adversarial networks, predator-prey interactions, protective colouration, signalling patterns | 241Methods in Ecology and Evoluঞon TALAS eT AL.

BackgroundIt is well established that there is anxiety-related variation between observers in the very earliest, pre-attentive stage of visual processing of images such as emotionally expressive faces, often leading to enhanced attention to threat in a variety of disorders and traits. Whether there is also variation in early-stage affective (i.e. valenced) responses resulting from such images, however, is not yet known. The present study used the subliminal affective priming paradigm to investigate whether people varying in trait social anxiety also differ in their affective responses to very briefly presented, emotionally expressive face images.Methodology/Principal FindingsParticipants (n = 67) completed a subliminal affective priming task, in which briefly presented and smiling, neutral and angry faces were shown for 10 ms durations (below objective and subjective thresholds for visual discrimination), and immediately followed by a randomly selected Chinese character mask (2000 ms). Ratings of participants' liking for each Chinese character indicated the degree of valenced affective response made to the unseen emotive images. Participants' ratings of their liking for the Chinese characters were significantly influenced by the type of face image preceding them, with smiling faces generating more positive ratings than neutral and angry ones (F(2,128) = 3.107, p<0.05). Self-reported social anxiety was positively correlated with ratings of smiling relative to neutral-face primed characters (Pearson's r = .323, p<0.01). Individual variation in self-reported mood awareness was not associated with ratings.ConclusionsTrait social anxiety is associated with individual variation in affective responding, even in response to the earliest, pre-attentive stage of visual image processing. However, the fact that these priming effects are limited to smiling and not angry (i.e. threatening) images leads us to propose that the pre-attentive processes involved in generating the subliminal affective priming effect may be different from those that generate attentional biases in anxious individuals.