Explore the vast range of titles available.

Computers can beat humans at increasingly complex games, including chess and Go. However, these programs are typically constructed for a particular game, exploiting its properties, such as the symmetries of the board on which it is played. Silver et al. developed a program called AlphaZero, which taught itself to play Go, chess, and shogi (a Japanese version of chess) (see the Editorial, and the Perspective by Campbell). AlphaZero managed to beat state-of-the-art programs specializing in these three games. The ability of AlphaZero to adapt to various game rules is a notable step toward achieving a general game-playing system. Science , this issue p. 1140 ; see also pp. 1087 and 1118 AlphaZero teaches itself to play three different board games and beats state-of-the-art programs in each. The game of chess is the longest-studied domain in the history of artificial intelligence. The strongest programs are based on a combination of sophisticated search techniques, domain-specific adaptations, and handcrafted evaluation functions that have been refined by human experts over several decades. By contrast, the AlphaGo Zero program recently achieved superhuman performance in the game of Go by reinforcement learning from self-play. In this paper, we generalize this approach into a single AlphaZero algorithm that can achieve superhuman performance in many challenging games. Starting from random play and given no domain knowledge except the game rules, AlphaZero convincingly defeated a world champion program in the games of chess and shogi (Japanese chess), as well as Go.

Building a large-scale quantum computer requires effective strategies to correct errors that inevitably arise in physical quantum systems 1 . Quantum error-correction codes 2 present a way to reach this goal by encoding logical information redundantly into many physical qubits. A key challenge in implementing such codes is accurately decoding noisy syndrome information extracted from redundancy checks to obtain the correct encoded logical information. Here we develop a recurrent, transformer-based neural network that learns to decode the surface code, the leading quantum error-correction code 3 . Our decoder outperforms other state-of-the-art decoders on real-world data from Google’s Sycamore quantum processor for distance-3 and distance-5 surface codes 4 . On distances up to 11, the decoder maintains its advantage on simulated data with realistic noise including cross-talk and leakage, utilizing soft readouts and leakage information. After training on approximate synthetic data, the decoder adapts to the more complex, but unknown, underlying error distribution by training on a limited budget of experimental samples. Our work illustrates the ability of machine learning to go beyond human-designed algorithms by learning from data directly, highlighting machine learning as a strong contender for decoding in quantum computers. A recurrent, transformer-based neural network, called AlphaQubit, learns high-accuracy error decoding to suppress the errors that occur in quantum systems, opening the prospect of using neural-network decoders for real quantum hardware.

Response inhibition refers to the suppression of prepared or initiated actions. Typically, the go/no-go task (GNGT) or the stop signal task (SST) are used interchangeably to capture individual differences in response inhibition. On the one hand, factor analytic and conjunction neuroimaging studies support the association of both tasks with a single inhibition construct. On the other hand, studies that directly compare the two tasks indicate distinct mechanisms, corresponding to action restraint and cancellation in the GNGT and SST, respectively. We addressed these contradictory findings with the aim to identify the core differences in the temporal dynamics of the functional networks that are recruited in both tasks. We extracted the time-courses of sensory, motor, attentional, and cognitive control networks by group independent component (G-ICA) analysis of electroencephalography (EEG) data from both tasks. Additionally, electromyography (EMG) from the responding effector muscles was recorded to detect the timing of response inhibition. The results indicated that inhibitory performance in the GNGT may be comparable to response selection mechanisms, reaching peripheral muscles at around 316 ​ms. In contrast, inhibitory performance in the SST is achieved via biasing of the sensorimotor system in preparation for stopping, followed by fast sensory, motor and frontal integration during outright stopping. Inhibition can be detected at the peripheral level at 140 ​ms after stop stimulus presentation. The GNGT and the SST therefore seem to recruit widely different neural dynamics, implying that the interchangeable use of superficially similar inhibition tasks in both basic and clinical research is unwarranted.

Knowing protein function is crucial to advance molecular and medical biology, yet experimental function annotations through the Gene Ontology (GO) exist for fewer than 0.5% of all known proteins. Computational methods bridge this sequence-annotation gap typically through homology-based annotation transfer by identifying sequence-similar proteins with known function or through prediction methods using evolutionary information. Here, we propose predicting GO terms through annotation transfer based on proximity of proteins in the SeqVec embedding rather than in sequence space. These embeddings originate from deep learned language models (LMs) for protein sequences (SeqVec) transferring the knowledge gained from predicting the next amino acid in 33 million protein sequences. Replicating the conditions of CAFA3, our method reaches an F max of 37 ± 2%, 50 ± 3%, and 57 ± 2% for BPO, MFO, and CCO, respectively. Numerically, this appears close to the top ten CAFA3 methods. When restricting the annotation transfer to proteins with < 20% pairwise sequence identity to the query, performance drops (F max BPO 33 ± 2%, MFO 43 ± 3%, CCO 53 ± 2%); this still outperforms naïve sequence-based transfer. Preliminary results from CAFA4 appear to confirm these findings. Overall, this new concept is likely to change the annotation of proteins, in particular for proteins from smaller families or proteins with intrinsically disordered regions.

Alzheimer's disease is a severe irreversible syndrome, characterized by a slow and progressive cognitive decline that interferes with the standard instrumental and essential functions of daily life. Promptly identifying the impairment of particular cognitive functions could be a fundamental condition to limit, through preventive or therapeutic interventions, the functional damages found in this degenerative dementia. This study aims to analyse, through a systematic review of the studies, the sensitivity of four experimental paradigms (Wisconsin Card Sorting Test, Stroop Task, Go/No-Go Task, and Flanker Task) considered as golden standard instruments for executive functions assessment in elderly subjects affected by Alzheimer dementia. This review was carried out according to the PRISMA method. Forty-five studies comparing the executive performance of patients with Alzheimer's dementia (diagnosed according to different classification criteria for dementia) and healthy elderly patients both over the age of sixty, were selected. For the research, PubMed, PsycINFO, PsycArticles databases were used. The study highlighted the importance of using standard protocols to evaluate executive dysfunction in Alzheimer's disease. The Stroop task allows discriminating better between healthy and pathological aging.

Excessive use of the internet is increasingly recognised as a global public health concern. Individual studies have reported cognitive impairment in problematic internet use (PIU), but have suffered from various methodological limitations. Confirmation of cognitive deficits in PIU would support the neurobiological plausibility of this disorder. To conduct a rigorous meta-analysis of cognitive performance in PIU from case-control studies; and to assess the impact of study quality, the main type of online behaviour (for example gaming) and other parameters on the findings. A systematic literature review was conducted of peer-reviewed case-controlled studies comparing cognition in people with PIU (broadly defined) with that of healthy controls. Findings were extracted and subjected to a meta-analysis where at least four publications existed for a given cognitive domain of interest. The meta-analysis comprised 2922 participants across 40 studies. Compared with controls, PIU was associated with significant impairment in inhibitory control (Stroop task Hedge's g = 0.53 (s.e. = 0.19-0.87), stop-signal task g = 0.42 (s.e. = 0.17-0.66), go/no-go task g = 0.51 (s.e. = 0.26-0.75)), decision-making (g = 0.49 (s.e. = 0.28-0.70)) and working memory (g = 0.40 (s.e. = 0.20-0.82)). Whether or not gaming was the predominant type of online behaviour did not significantly moderate the observed cognitive effects; nor did age, gender, geographical area of reporting or the presence of comorbidities. PIU is associated with decrements across a range of neuropsychological domains, irrespective of geographical location, supporting its cross-cultural and biological validity. These findings also suggest a common neurobiological vulnerability across PIU behaviours, including gaming, rather than a dissimilar neurocognitive profile for internet gaming disorder. S.R.C. consults for Cambridge Cognition and Shire. K.I.'s research activities were supported by Health Education East of England Higher Training Special interest sessions. A.E.G.'s research has been funded by Innovational grant (VIDI-scheme) from ZonMW: (91713354). N.A.F. has received research support from Lundbeck, Glaxo-SmithKline, European College of Neuropsychopharmacology (ECNP), Servier, Cephalon, Astra Zeneca, Medical Research Council (UK), National Institute for Health Research, Wellcome Foundation, University of Hertfordshire, EU (FP7) and Shire. N.A.F. has received honoraria for lectures at scientific meetings from Abbott, Otsuka, Lundbeck, Servier, Astra Zeneca, Jazz pharmaceuticals, Bristol Myers Squibb, UK College of Mental Health Pharmacists and British Association for Psychopharmacology (BAP). N.A.F. has received financial support to attend scientific meetings from RANZCP, Shire, Janssen, Lundbeck, Servier, Novartis, Bristol Myers Squibb, Cephalon, International College of Obsessive-Compulsive Spectrum Disorders, International Society for Behavioral Addiction, CINP, IFMAD, ECNP, BAP, the World Health Organization and the Royal College of Psychiatrists. N.A.F. has received financial royalties for publications from Oxford University Press and payment for editorial duties from Taylor and Francis. J.E.G. reports grants from the National Center for Responsible Gaming, Forest Pharmaceuticals, Takeda, Brainsway, and Roche and others from Oxford Press, Norton, McGraw-Hill and American Psychiatric Publishing outside of the submitted work.

Depression is characterised by a heightened self-focus, which is believed to be associated with differences in emotion and reward processing. However, the precise relationship between these cognitive domains is not well understood. We examined the role of self-reference in emotion and reward processing, separately and in combination, in relation to depression. Adults experiencing varying levels of depression ( = 144) completed self-report depression measures (PHQ-9, BDI-II). We measured self, emotion and reward processing, separately and in combination, using three cognitive tasks. When self-processing was measured independently of emotion and reward, in a simple associative learning task, there was little association with depression. However, when self and emotion processing occurred in combination in a self-esteem go/no-go task, depression was associated with an increased positive other bias [ = 3.51, 95% confidence interval (CI) 1.24-5.79]. When the self was processed in relation to emotion and reward, in a social evaluation learning task, depression was associated with reduced positive self-biases ( = 0.11, 95% CI 0.05-0.17). Depression was associated with enhanced positive implicit associations with others, and reduced positive learning about the self, culminating in reduced self-favouring biases. However, when self, emotion and reward processing occurred independently there was little evidence of an association with depression. Treatments targeting reduced positive self-biases may provide more sensitive targets for therapeutic intervention and potential biomarkers of treatment responses, allowing the development of more effective interventions.

The orbitofrontal cortex (OFC) encodes expected outcomes and plays a critical role in flexible, outcome-guided behavior. The OFC projects to primary visual cortex (V1), yet the function of this top-down projection is unclear. We find that optogenetic activation of OFC projection to V1 reduces the amplitude of V1 visual responses via the recruitment of local somatostatin-expressing (SST) interneurons. Using mice performing a Go/No-Go visual task, we show that the OFC projection to V1 mediates the outcome-expectancy modulation of V1 responses to the reward-irrelevant No-Go stimulus. Furthermore, V1-projecting OFC neurons reduce firing during expectation of reward. In addition, chronic optogenetic inactivation of OFC projection to V1 impairs, whereas chronic activation of SST interneurons in V1 improves the learning of Go/No-Go visual task, without affecting the immediate performance. Thus, OFC top-down projection to V1 is crucial to drive visual associative learning by modulating the response gain of V1 neurons to non-relevant stimulus. The orbitofrontal cortex (OFC) encodes expected outcomes and plays a key role in outcome-guided behavior. The authors show here that the top-down projection from the OFC to the visual cortex drives visual associative learning by modulating the response gain of V1 neurons to non-relevant stimuli.

•Teaching is a critical component of our identity as surgeons.•The skills needed to be an effective teacher of surgery can be learned.•Teaching is how we connect with our learners and the surgical community.