In comparison with most animal behaviours, circadian rhythms have a well-characterized molecular genetic basis. Detailed studies of circadian clock genes in 'model' organisms provide a foundation for interpreting the functional and evolutionary significance of polymorphic circadian clock genes found within free-living animal populations. Here, we describe allelic variation in a region of the avian Clock orthologue which encodes a functionally significant polyglutamine repeat (ClkpolyQcds), within free-living populations of two passerine birds, the migratory bluethroat (Luscinia svecica) and the predominantly nonmigratory blue tit (Cyanistes caeruleus). Multiple ClkpolyQcds alleles were found within populations of both species (bluethroat: 12 populations, 7 alleles; blue tit: 14 populations, 9 alleles). Some populations of both species were differentiated at the ClkpolyQcds locus as measured by F(ST) and R(ST) values. Among the blue tit, but not bluethroat populations, we found evidence of latitudinal clines in (i) mean ClkpolyQcds repeat length, and (ii) the proportions of three ClkpolyQcds genotype groupings. Parallel analyses of microsatellite allele frequencies, which are considered to reflect selectively neutral processes, indicate that interpopulation allele frequency variation at the ClkpolyQcds and microsatellite loci does not reflect the same underlying demographic processes. The possibility that the observed interpopulation ClkpolyQcds allele frequency variation is, at least in part, maintained by selection for microevolutionary adaptation to photoperiodic parameters correlated with latitude warrants further study.

Females should prefer to be fertilized by males that increase the genetic quality of their offspring. In vertebrates, genes of the major histocompatibility complex (MHC) play a key role in the acquired immune response and have been shown to affect mating preferences. They are therefore important candidates for the link between mate choice and indirect genetic benefits. Higher MHC diversity may be advantageous because this allows a wider range of pathogens to be detected and combated. Furthermore, individuals harbouring rare MHC alleles might better resist pathogen variants that have evolved to evade common MHC alleles. In the Seychelles warbler, females paired with low MHC-diversity males elevate the MHC diversity of their offspring to levels comparable to the population mean by gaining extra-pair fertilizations. Here, we investigate whether increased MHC diversity results in higher life expectancy and whether there are any additional benefits of extra-pair fertilizations. Our 10-year study found a positive association between MHC diversity and juvenile survival, but no additional survival advantage of extra-pair fertilizations. In addition, offspring with a specific allele (Ase-ua4) had a fivefold longer life expectancy than offspring without this allele. Consequently, the interacting effects of sexual selection and pathogen-mediated viability selection appear to be important for maintaining MHC variation in the Seychelles warbler. Our study supports the prediction that MHC-dependent extra-pair fertilizations result in genetic benefits for offspring in natural populations. However, such genetic benefits might be hidden and not necessarily apparent in the widely used fitness comparison of extra- and within-pair offspring.

ordered alphabetically, † Equal contributions, ordered alphabetically, ‡ Equal senior contributions Building models that can be rapidly adapted to numerous tasks using only a handful of annotated examples is an open challenge for multimodal machine learning research. We introduce Flamingo, a family of Visual Language Models (VLM) with this ability. Flamingo models include key architectural innovations to: (i) bridge powerful pretrained vision-only and language-only models, (ii) handle sequences of arbitrarily interleaved visual and textual data, and (iii) seamlessly ingest images or videos as inputs. Thanks to their flexibility, Flamingo models can be trained on large-scale multimodal web corpora containing arbitrarily interleaved text and images, which is key to endow them with in-context few-shot learning capabilities. We perform a thorough evaluation of the proposed Flamingo models, exploring and measuring their ability to rapidly adapt to a variety of image and video understanding benchmarks. These include open-ended tasks such as visual question-answering, where the model is prompted with a question which it has to answer, captioning tasks, which evaluate the ability to describe a scene or an event, and close-ended tasks such as multiple choice visual question-answering. For tasks lying anywhere on this spectrum, we demonstrate that a single Flamingo model can achieve a new state of the art for few-shot learning, simply by prompting the model with task-specific examples. On many of these benchmarks, Flamingo actually surpasses the performance of models that are fine-tuned on thousands of times more task-specific data.

A common vision from science fiction is that robots will one day inhabit our physical spaces, sense the world as we do, assist our physical labours, and communicate with us through natural language. Here we study how to design artificial agents that can interact naturally with humans using the simplification of a virtual environment. This setting nevertheless integrates a number of the central challenges of artificial intelligence (AI) research: complex visual perception and goal-directed physical control, grounded language comprehension and production, and multi-agent social interaction. To build agents that can robustly interact with humans, we would ideally train them while they interact with humans. However, this is presently impractical. Therefore, we approximate the role of the human with another learned agent, and use ideas from inverse reinforcement learning to reduce the disparities between human-human and agent-agent interactive behaviour. Rigorously evaluating our agents poses a great challenge, so we develop a variety of behavioural tests, including evaluation by humans who watch videos of agents or interact directly with them. These evaluations convincingly demonstrate that interactive training and auxiliary losses improve agent behaviour beyond what is achieved by supervised learning of actions alone. Further, we demonstrate that agent capabilities generalise beyond literal experiences in the dataset. Finally, we train evaluation models whose ratings of agents agree well with human judgement, thus permitting the evaluation of new agent models without additional effort. Taken together, our results in this virtual environment provide evidence that large-scale human behavioural imitation is a promising tool to create intelligent, interactive agents, and the challenge of reliably evaluating such agents is possible to surmount. See videos for an overview of the manuscript, training time-lapse, and human-agent interactions.