RESUMO
BACKGROUND: Interlimb transfer of sequential motor learning (SML) refers to the positive influence of prior experiences in performing the same sequential movements using different effectors. Despite evidence from intermanual SML, and while most daily living activities involve interlimb cooperation and coordination between the four limbs, nothing is known about bilateral SML transfer between the upper and lower limbs. RESEARCH QUESTION: We examined the transfer of bilateral SML from the upper to the lower limbs and vice versa. METHODS: Twenty-four participants had to learn an initial bilateral SML task using the upper limbs and then performed the same sequence using the lower limbs during a transfer SML task. They performed the reversed situation 1 month apart. The performance was evaluated at the beginning and the end of both initial and transfer SML practice phases. RESULTS: Significant and reciprocal transfer gains in performance were observed regardless of the effectors. Greater transfer gains in performance were observed at the beginning of the transfer SML from the lower to the upper limbs (44â¯%) but these gains vanished after practice with the transfer effectors (5â¯%). Although smaller gains were initially achieved in the transfer of SML from the upper to the lower limbs (15â¯%), these gains persisted and remained significant (9â¯%) after practice with the transfer effectors. SIGNIFICANCE: Our results provide evidence of a reciprocal and asymmetrical interlimb transfer of bilateral SML between the upper and lower limbs. These findings could be leveraged as a relevant strategy in the context of sports and functional rehabilitation.
RESUMO
Deep neural networks have made tremendous gains in emulating human-like intelligence, and have been used increasingly as ways of understanding how the brain may solve the complex computational problems on which this relies. However, these still fall short of, and therefore fail to provide insight into how the brain supports strong forms of generalization of which humans are capable. One such case is out-of-distribution (OOD) generalization - successful performance on test examples that lie outside the distribution of the training set. Here, we identify properties of processing in the brain that may contribute to this ability. We describe a two-part algorithm that draws on speciï¬c features of neural computation to achieve OOD generalization, and provide a proof of concept by evaluating performance on two challenging cognitive tasks. First we draw on the fact that the mammalian brain represents metric spaces using grid cell code (e.g., in the entorhinal cortex): abstract representations of relational structure, organized in recurring motifs that cover the representational space. Second, we propose an attentional mechanism that operates over the grid cell code using determinantal point process (DPP), that we call DPP attention (DPP-A) - a transformation that ensures maximum sparseness in the coverage of that space. We show that a loss function that combines standard task-optimized error with DPP-A can exploit the recurring motifs in the grid cell code, and can be integrated with common architectures to achieve strong OOD generalization performance on analogy and arithmetic tasks. This provides both an interpretation of how the grid cell code in the mammalian brain may contribute to generalization performance, and at the same time a potential means for improving such capabilities in artiï¬cial neural networks.
Assuntos
Células de Grade , Redes Neurais de Computação , Humanos , Células de Grade/fisiologia , Algoritmos , Modelos Neurológicos , Animais , Atenção/fisiologia , Encéfalo/fisiologia , Córtex Entorrinal/fisiologiaRESUMO
Introduction: Listeners can rapidly adapt to English speech produced by non-native speakers of English with unfamiliar accents. Prior work has shown that the type and number of talkers contained within a stimulus set may impact rate and magnitude of learning, as well as any generalization of learning. However, findings across the literature have been inconsistent, with relatively little study of these effects in populations of older listeners. Methods: In this study, adaptation and generalization to unfamiliar talkers with familiar and unfamiliar accents are studied in younger normal-hearing adults and older adults with and without hearing loss. Rate and magnitude of adaptation are modelled using both generalized linear mixed effects regression and generalized additive mixed effects modelling. Results: Rate and magnitude of adaptation were not impacted by increasing the number of talkers and/or varying the consistency of non-native English accents across talkers. Increasing the number of talkers did strengthen generalization of learning for a talker with a familiar non-native accent, but not for an unfamiliar accent. Aging alone did not diminish adaptation or generalization. Discussion: These findings support prior evidence of a limited benefit for talker variability in facilitating generalization of learning for non-native accented speech, and extend the findings to older adults.
RESUMO
Domain Generalization (DG) focuses on the Out-Of-Distribution (OOD) generalization, which is able to learn a robust model that generalizes the knowledge acquired from the source domain to the unseen target domain. However, due to the existence of the domain shift, domain-invariant representation learning is challenging. Guided by fine-grained knowledge, we propose a novel paradigm Mask-Shift-Inference (MSI) for DG based on the architecture of Convolutional Neural Networks (CNN). Different from relying on a series of constraints and assumptions for model optimization, this paradigm novelly shifts the focus to feature channels in the latent space for domain-invariant representation learning. We put forward a two-branch working mode of a main module and multiple domain-specific sub-modules. The latter can only achieve good prediction performance in its own specific domain but poor predictions in other source domains, which provides the main module with the fine-grained knowledge guidance and contributes to the improvement of the cognitive ability of MSI. Firstly, during the forward propagation of the main module, the proposed MSI accurately discards unstable channels based on spurious classifications varying across domains, which have domain-specific prediction limitations and are not conducive to generalization. In this process, a progressive scheme is adopted to adaptively increase the masking ratio according to the training progress to further reduce the risk of overfitting. Subsequently, our paradigm enters the compatible shifting stage before the formal prediction. Based on maximizing semantic retention, we implement the domain style matching and shifting through the simple transformation through Fourier transform, which can explicitly and safely shift the target domain back to the source domain whose style is closest to it, requiring no additional model updates and reducing the domain gap. Eventually, the paradigm MSI enters the formal inference stage. The updated target domain is predicted in the main module trained in the previous stage with the benefit of familiar knowledge from the nearest source domain masking scheme. Our paradigm is logically progressive, which can intuitively exclude the confounding influence of domain-specific spurious information along with mitigating domain shifts and implicitly perform semantically invariant representation learning, achieving robust OOD generalization. Extensive experimental results on PACS, VLCS, Office-Home and DomainNet datasets verify the superiority and effectiveness of the proposed method.
RESUMO
The gestures we produce serve a variety of functions-they affect our communication, guide our attention and help us think and change the way we think. Gestures can consequently also help us learn, generalize what we learn and retain that knowledge over time. The effects of gesture-based instruction in mathematics have been well studied. However, few of these studies are directly applicable to classroom environments. Here, we review literature that highlights the benefits of producing and observing gestures when teaching and learning mathematics, and we provide suggestions for designing research studies with an eye towards how gestures can feasibly be applied to classroom learning. This article is part of the theme issue 'Minds in movement: embodied cognition in the age of artificial intelligence'.
Assuntos
Gestos , Aprendizagem , Matemática , Humanos , Criança , Matemática/educação , Ensino , Professores Escolares/psicologia , Cognição , Instituições AcadêmicasRESUMO
A good theory of mathematical beauty is more practical than any current observation, as new predictions about physical reality can be self-consistently verified. This belief applies to the current status of understanding deep neural networks including large language models and even the biological intelligence. Toy models provide a metaphor of physical reality, allowing mathematically formulating the reality (i.e., the so-called theory), which can be updated as more conjectures are justified or refuted. One does not need to present all details in a model, but rather, more abstract models are constructed, as complex systems such as the brains or deep networks have many sloppy dimensions but much less stiff dimensions that strongly impact macroscopic observables. This type of bottom-up mechanistic modeling is still promising in the modern era of understanding the natural or artificial intelligence. Here, we shed light on eight challenges in developing theory of intelligence following this theoretical paradigm. Theses challenges are representation learning, generalization, adversarial robustness, continual learning, causal learning, internal model of the brain, next-token prediction, and the mechanics of subjective experience.
RESUMO
Faces can acquire emotional meaning by learning to associate individuals with specific behaviors. Here, we investigated emotional evaluation and brain activations toward faces of persons who had given negative or positive evaluations to others. Furthermore, we investigated how emotional evaluations and brain activation generalize to perceptually similar faces. Valence ratings indicated learning and generalization effects for both positive and negative faces. Brain activation, measured with functional magnetic resonance imaging (fMRI), showed significantly increased activation in the fusiform gyrus (FG) to negatively associated faces but not positively associated ones. Remarkably, brain activation in FG to faces to which emotional meaning (negative and positive) was successfully generalized was decreased compared to neutral faces. This suggests that the emotional relevance of faces is not simply associated with increased brain activation in visual areas. While, at least for negative conditions, faces paired with negative feedback behavior are related to potentiated brain responses, the opposite is seen for perceptually very similar faces despite generalized emotional responses.
RESUMO
Generalization, the tendency to respond in the same way to different but similar stimuli, is one of the main cognitive abilities that make category formation possible and thus is a prerequisite for efficiency in learning. Individuals with autism spectrum disorder (ASD) experience pervasive difficulty with producing generalized responses across materials, people, places, and contexts. Increasing evidence suggests that "ASD-like" social impairments appear endogenously and spontaneously in family dogs providing a high-validity model for understanding the phenotypic expression of human ASD. The present study aims to further investigate the dog model of ASD by the approach of searching for analogues in dogs showing "ASD-like" social impairments of cognitive phenomena in humans specific to ASD, specifically impairments of generalization abilities. We have tested 18 family dogs with formerly established "ASD-like" behaviour scores (F1, F2, F3) in a generalization task involving three conditions (size, colour and texture). We found a significant association between F1 scores and test performance as well as improvement during testing sessions. Our study provides further support for the notion that dogs with lower social competence-similarly to humans with ASD-exhibit attentional and perceptual abnormalities, such as being sensitive to minor changes to a non-adaptive extent.
Assuntos
Transtorno do Espectro Autista , Comportamento Animal , Comportamento Social , Animais , Cães , Transtorno do Espectro Autista/fisiopatologia , Transtorno do Espectro Autista/psicologia , Masculino , Feminino , Generalização do Estímulo , Humanos , Modelos Animais de DoençasRESUMO
Psychological interventions tend to be confrontational in nature. That is, when psychologists seek to bring about change in beliefs, attitudes, or behaviors, they often do so by directly confronting the presumed barrier to change. Confrontational approaches can be effective, but suffer from limitations to their efficacy, such as the possibility of arousing discomfort or defensiveness from the recipient. The current piece seeks to highlight an alternative strategy that we refer to as bypassing, which refers to a general approach for bringing about behavior change without confrontation. Leveraging insights from research on misinformation, stereotypes, and persuasion, we present evidence that non-confrontational approaches can be as effective, if not more so, than the traditional confrontational paradigm.
RESUMO
Foot-shock paradigms have provided valuable insights into the neurobiology of stress and fear conditioning. An extensive body of literature indicates that shock exposure can elicit both conditioned and unconditioned effects, although delineating between the two is a challenging task. This distinction holds crucial implications not only for the theoretical interpretation of fear conditioning, but also for properly evaluating putative preclinical models of post-traumatic stress disorder (PTSD) involving shock exposure. The characteristics of shocks (intensity and number) affect the strength of learning, but how these characteristics interact to influence conditioned and unconditioned consequences of shocks are poorly known. In this study, we aimed to investigate in adult male rats the impact of varying shock number and intensity on the endocrine and behavioral response to contextual fear conditioning and fear generalization to a novel environment markedly distinct from the shock context (i.e., fear generalization). Classical biological markers of stress (i.e., ACTH, corticosterone, and prolactin) were sensitive to manipulations of shock parameters, whereas these parameters had a limited effect on contextual fear conditioning (evaluated by freezing and distance traveled). In contrast, behavior in different novel contexts (fear generalization) was specifically sensitive to shock intensity. Notably, altered behavior in novel contexts markedly improved, but not completely normalized after fear extinction, hypoactivity apparently being the result of both conditioned and unconditioned effects of foot-shock exposure. The present results will contribute to a better understanding of shock exposure as a putative animal model of PTSD.
RESUMO
RATIONALE: Maladaptive fear responses, including sensitized threat reactions and overgeneralization, contribute to anxiety disorders such as generalized anxiety disorder and post-traumatic stress disorder. Although stress intensity influences the generation and extent of these maladaptive fears, the underlying mechanisms remain unclear. OBJECTIVES: The present study examined whether varying footshock stress intensity and inhibition of protein synthesis have differential effect on fear sensitization and generalization in mice. METHODS: Mice were subjected to a classic fear conditioning protocol involving five different levels of footshock intensities. Prior to fear acquisition, the protein synthesis inhibitor cycloheximide (CHX) was administered intraperitoneally. Fear sensitization to white noise and fear generalization to tones with frequencies differing from the conditioned tone were assessed at either 2 or 4 days after fear acquisition. RESULTS: The results showed that, although varying shock intensities (except the lowest) led to a similar pattern of increased freezing during auditory cues in fear acquisition, the extent of both fear sensitization and generalization increased with the intensity of the footshock in the following days. As shock intensities increased, there was a proportional rise in sensitized fear to white noise and generalized freezing to tones with frequencies progressively closer to the conditioned stimulus. Mildest shocks did not induce discriminative conditioned fear memory, whereas the most intense shocks led to pronounced fear generalization. Administration of CHX before fear acquisition did not affect sensitized fear but reduced generalization of freezing to tones dissimilar from the conditioned stimulus in the group exposed to the most intense shock. CONCLUSIONS: Our results suggest that maladaptive fear responses elicited by varying stress intensities exhibit distinct characteristics. The effect of CHX to prevent overgeneralization without affecting discriminative fear memory points to potential therapeutic approaches for fear-related disorders, suggesting the possibility of mitigating overgeneralization while preserving necessary fear discrimination.
RESUMO
OBJECTIVE: We recently demonstrated that learning abilities among school-age children vary following frequency discrimination (FD) training, with some exhibiting mature adult-like learning while others performing poorly (non-adult-like learners). This study tested the hypothesis that children's post-training generalisation is related to their learning maturity. Additionally, it investigated how training duration influences children's generalisation, considering the observed decrease with increased training in adults. DESIGN: Generalisation to the untrained ear and untrained 2000 Hz frequency was assessed following single-session or nine-session 1000 Hz FD training, using an adaptive forced-choice procedure. Two additional groups served as controls for the untrained frequency. STUDY SAMPLE: Fifty-four children aged 7-9 years and 59 adults aged 18-30 years. RESULTS: (1) Only adult-like learners generalised their learning gains across frequency or ear, albeit less efficiently than adults; (2) As training duration increased children experienced reduced generalisation, similar to adults; (3) Children's performance in the untrained tasks correlated strongly with their trained task performance after the first training session. CONCLUSIONS: Auditory skill learning and its generalisation do not necessarily mature contemporaneously, although mature learning is a prerequisite for mature generalisation. Furthermore, in children, as in adults, more practice makes rather specific experts. These findings should be considered when designing training programs.
RESUMO
As genomic selection emerges as a promising breeding method for both plants and animals, numerous methods have been introduced and applied to various real and simulated data sets. Research suggests that no single method is universally better than others; rather, performance is highly dependent on the characteristics of the data and the nature of the prediction task. This implies that each method has its strengths and weaknesses. In this study, we exploit this notion and propose a different approach. Rather than comparing multiple methods to determine the best one for a particular study, we advocate combining multiple methods to achieve better performance than each method in isolation. In pursuit of this goal, we introduce and develop a computational method of the stacked generalization within ensemble methods. In this method, the meta-model merges predictions from multiple base models to achieve improved performance. We applied this method to plant and animal data and compared its performance with currently available methods using standard performance metrics. We found that the proposed method yielded a lower or comparable mean squared error in predicting phenotypes compared to the current methods. In addition, the proposed method showed greater resistance to overfitting compared to the current methods. Further analysis included statistical hypothesis testing, which showed that the proposed method outperformed or matched the current methods. In summary, the proposed stacked generalization integrates currently available methods to achieve stable and better performance. In this context, our study provides general recommendations for effective practices in genomic selection.
RESUMO
Practicing complex locomotor skills, such as those involving a step sequence engages distinct perceptual and motor mechanisms that support the recall of learning under new conditions (i.e., skill transfer). While sleep has been shown to enhance learning of sequences of fine movements (i.e., sleep-dependent consolidation), here we examined whether this benefit extends to learning of a locomotor pattern. Specifically, we tested the perceptual and motor learning of a locomotor sequence following sleep compared to wake. We hypothesized that post-practice sleep would increase locomotor sequence learning in the perceptual, but not in the motor domain. In this study, healthy young adult participants (n = 48; 18-33 years) practiced a step length sequence on a treadmill cued by visual stimuli displayed on a screen during training. Participants were then tested in a perceptual condition (backward walking with the same visual stimuli), or a motor condition (forward walking but with an inverted screen). Skill was assessed immediately, and again after a 12-h delay following overnight sleep or daytime wake (n = 12 for each interval/condition). Off-line learning improved following sleep compared to wake, but only for the perceptual condition. Our results suggest that perceptual and motor sequence learning are processed separately after locomotor training, and further points to a benefit of sleep that is rooted in the perceptual as opposed to the motor aspects of motor learning.
Assuntos
Aprendizagem , Sono , Humanos , Adulto , Sono/fisiologia , Masculino , Feminino , Adulto Jovem , Aprendizagem/fisiologia , Adolescente , Destreza Motora/fisiologia , Locomoção/fisiologia , Movimento/fisiologia , Desempenho Psicomotor/fisiologia , Caminhada/fisiologiaRESUMO
Evaluation metrics for prediction error, model selection and model averaging on space-time data are understudied and poorly understood. The absence of independent replication makes prediction ambiguous as a concept and renders evaluation procedures developed for independent data inappropriate for most space-time prediction problems. Motivated by air pollution data collected during California wildfires in 2008, this manuscript attempts a formalization of the true prediction error associated with spatial interpolation. We investigate a variety of cross-validation (CV) procedures employing both simulations and case studies to provide insight into the nature of the estimand targeted by alternative data partition strategies. Consistent with recent best practice, we find that location-based cross-validation is appropriate for estimating spatial interpolation error as in our analysis of the California wildfire data. Interestingly, commonly held notions of bias-variance trade-off of CV fold size do not trivially apply to dependent data, and we recommend leave-one-location-out (LOLO) CV as the preferred prediction error metric for spatial interpolation.
RESUMO
Aim: Deep learning's widespread use prompts heightened scrutiny, particularly in the biomedical fields, with a specific focus on model generalizability. This study delves into the influence of training data characteristics on the generalization performance of models, specifically in cardiac abnormality detection. Materials & methods: Leveraging diverse electrocardiogram datasets, models are trained on subsets with varying characteristics and subsequently compared for performance. Additionally, the introduction of the attention mechanism aims to improve generalizability. Results: Experiments reveal that using a balanced dataset, just 1% of a large dataset, leads to equal performance in generalization tasks, notably in detecting cardiology abnormalities. Conclusion: This balanced training data notably enhances model generalizability, while the integration of the attention mechanism further refines the model's ability to generalize effectively.
This study tackles a common problem for deep learning models: they often struggle when faced with new, unfamiliar data that they have not been trained on. This phenomenon is also known as performance drop in out-of-distribution generalization. This reduced performance on out-of-distribution generalization is a key focus of the research, aiming to improve the models' ability to handle diverse data sets beyond their training data.The study examines how the characteristics of the dataset used to train deep learning models affect their ability to detect abnormal heart activities when applied to new, unseen data. Researchers trained these models using various sets of electrocardiogram (ECG) data and then evaluated their performance in identifying abnormalities. They also introduced an attention mechanism to enhance the models' learning capabilities. The attention mechanism in deep learning is like a spotlight that helps the model focus on important information while ignoring less relevant details.The findings were particularly noteworthy. Despite being trained on a small, well-balanced subset of a larger dataset, the models excelled in detecting heart abnormalities in new, unfamiliar data. This training method significantly improved the models' generalization and performance with unseen data. Furthermore, integrating the attention mechanism substantially enhanced the models' ability to generalize effectively on new information.
Assuntos
Aprendizado Profundo , Eletrocardiografia , Humanos , Eletrocardiografia/métodosRESUMO
PREMISE: Under pollinator limitations, specialized pollination syndromes may evolve toward contrasting responses: a generalized syndrome with increased pollinator attraction, pollinator reward, and pollen transfer capacity; or the selfing syndrome with increased self-pollen deposition, but reduced pollinator attraction and pollen transfer capacity. The buzz-pollination syndrome is specialized to explore female vibrating bees as pollinators. However, vibrating bees become less-active pollinators at montane areas of the Atlantic Forest (AF) domain. This study investigated whether the specialized buzz-pollination syndrome would evolve toward an alternative floral syndrome in montane areas of the AF domain, considering a generalized and the selfing syndromes as alternative responses. METHODS: We utilized a lineage within the buzz-pollinated Miconia as study system, contrasting floral traits between montane AF-endemic and non-endemic species. We measured and validated floral traits that were proxies for pollinator attraction, reward access, pollen transfer capacity, and self-pollen deposition. We inferred the evolution of floral trait via phylogenetic comparative methods. RESULTS: AF-endemic species have selectively evolved greater reward access and more frequently had generalist pollination. Nonetheless, AF-endemic species also have selectively evolved toward lower pollen transfer capacity and greater self pollination. These patterns indicated a complex evolutionary process that has jointly favored a generalized and the selfing syndromes. CONCLUSIONS: The buzz pollination syndrome can undergo an evolutionary disruption in montane areas of the AF domain. This floral syndrome is likely more labile than often assumed, allowing buzz-pollinated plants to reproduce in environments where vibrating bees are less-reliable pollinators.
Assuntos
Evolução Biológica , Flores , Polinização , Animais , Abelhas/fisiologia , Flores/fisiologia , Filogenia , Pólen/fisiologiaRESUMO
OBJECTIVE: The ability to apply results from a study to a broader population remains a primary objective in translational science. Distinct from intrinsic elements of scientific rigor, the extrinsic concept of generalization requires there be alignment between a study cohort and population in which results are expected to be applied. Widespread efforts have been made to quantify representativeness of study cohorts. These techniques, however, often consider the study and target cohorts as monolithic collections that can be directly compared. Overlooking known impacts to health from socio-demographic and environmental factors tied to individual's geographical location, and potentially obfuscating misalignment in underrepresented population subgroups. This manuscript introduces several measures to account for geographic information in the assessment of cohort representation. METHODS: Metrics were defined across two themes: First, measures of recruitment, to assess if a study cohort is drawn at an expected rate and in an expected geographical pattern with respect to individuals in a reference cohort. Second, measures of individual characteristics, to assess if the individuals in the study cohort accurately reflect the sociodemographic, clinical, and geographic diversity observed across a reference cohort while accounting for the geospatial proximity of individuals. RESULTS: As an empirical demonstration, methods are applied to an active clinical study examining asthma in Black/African American patients at a US Midwestern pediatric hospital. Results illustrate how areas of over- and under-recruitment can be identified and contextualized in light of study recruitment patterns at an individual-level, highlighting the ability to identify a subset of features for which the study cohort closely resembled the broader population. In addition they provide an opportunity to dive deeper into misalignments, to identify study cohort members that are in some way distinct from the communities for which they are expected to represent. CONCLUSION: Together, these metrics provide a comprehensive spatial assessment of a study cohort with respect to a broader target population. Such an approach offers researchers a toolset by which to target expected generalization of results derived from a given study.
RESUMO
Visual reinforcement learning is important in various practical applications, such as video games, robotic manipulation, and autonomous navigation. However, a major challenge in visual reinforcement learning is the generalization to unseen environments, that is, how agents manage environments with previously unseen backgrounds. This issue is triggered mainly by the high unpredictability inherent in high-dimensional observation space. To deal with this problem, techniques including domain randomization and data augmentation have been explored; nevertheless, these methods still cannot attain a satisfactory result. This paper proposes a new method named Internal States Simulation Auxiliary (ISSA), which uses internal states to improve generalization in visual reinforcement learning tasks. Our method contains two agents, a teacher agent and a student agent: the teacher agent has the ability to directly access the environment's internal states and is used to facilitate the student agent's training; the student agent receives initial guidance from the teacher agent and subsequently continues to learn independently. From another perspective, our method can be divided into two phases, the transfer learning phase and traditional visual reinforcement learning phase. In the first phase, the teacher agent interacts with environments and imparts knowledge to the vision-based student agent. With the guidance of the teacher agent, the student agent is able to discover more effective visual representations that address the high unpredictability of high-dimensional observation space. In the next phase, the student agent autonomously learns from the visual information in the environment, and ultimately, it becomes a vision-based reinforcement learning agent with enhanced generalization. The effectiveness of our method is evaluated using the DMControl Generalization Benchmark and the DrawerWorld with texture distortions. Preliminary results indicate that our method significantly improves generalization ability and performance in complex continuous control tasks.