RESUMO
This review proposes a biologically plausible working model for the relationship between the 24-h activity cycle (24-HAC) and cardiovascular disease. The 24-HAC encompasses moderate-to-vigorous physical activity (MVPA), light physical activity, sedentary behavior (SB), and sleep. MVPA confers the greatest relative cardioprotective effect, when considering MVPA represents just 2% of the day if physical activity guidelines (30 min/day) are met. While we have well-established guidelines for MVPA, those for the remaining activity behaviors are vague. The vague guidelines are attributable to our limited mechanistic understanding of the independent and additive effects of these behaviors on the cardiovascular system. Our proposed biological model places arterial stiffness, a measure of vascular aging, as the key intermediate outcome. Starting with prolonged exposure to SB or static standing, we propose that the reported transient increases in arterial stiffness are driven by a cascade of negative hemodynamic effects following venous pooling. The subsequent autonomic, metabolic, and hormonal changes further impair vascular function. Vascular dysfunction can be offset by using mechanistic-informed interruption strategies and by engaging in protective behaviors throughout the day. Physical activity, especially MVPA, can confer protection by chronically improving endothelial function and associated protective mechanisms. Conversely, poor sleep, especially in duration and quality, negatively affects hormonal, metabolic, autonomic, and hemodynamic variables that can confound the physiological responses to next-day activity behaviors. Our hope is that the proposed biologically plausible working model will assist in furthering our understanding of the effects of these complex, interrelated activity behaviors on the cardiovascular system.
Assuntos
Sistema Cardiovascular , Rigidez Vascular , Ciclos de Atividade , Exercício Físico/fisiologia , Sono/fisiologia , AcelerometriaRESUMO
OBJECTIVES: The objective of this research was to develop and test methods for accessing and evaluating information on the biological plausibility of observed associations between exposures or interventions and outcomes to generate scientific evidence for action consistent with practice in systematic reviews. STUDY DESIGN: To undertake this research, we used the example of the observed associations between antimicrobial use in food animals and increased risks of human exposures to antimicrobial-resistant pathogens of zoonotic origin. METHODS: We conducted a scoping search using terms related to biological plausibility or mechanism to identify key references. As recommended by these references, we also used expert consultation with researchers and a public health informationist. We used their recommendations, which included expert consultation, to identify mechanisms relevant to biological plausibility of the association we selected to test. We used the reviews conducted by the World Health Organization (WHO) Guidelines Development Group in support of reducing antimicrobial use in food animal production to populate our model for assessing biological plausibility. RESULTS: We were able to develop a transparent model for biological plausibility based on the adverse outcome pathway used in toxicology and ecology. We were also able to populate this model using the WHO reviews. CONCLUSIONS: This analysis of biological plausibility used transparent and validated methods to assess the evidence used in systematic reviews based on the observational studies accessed through searches of the scientific literature. Given the importance of this topic in systematic reviews and evidence-based decision-making, further research is needed to define and test the methodological approaches to access and properly evaluate information from the scientific literature.
Assuntos
Produtos Biológicos/efeitos adversos , Modelos Biológicos , Saúde Pública , Revisões Sistemáticas como Assunto , Animais , Anti-Infecciosos/efeitos adversos , Anti-Infecciosos/uso terapêutico , Resistência Microbiana a Medicamentos , Doenças Transmitidas por Alimentos/epidemiologia , HumanosRESUMO
Clinical decisions in reproductive medicine are often made in uncertainty. To reduce uncertainty and to improve clinical decision-making, RCTs are increasingly called upon. A key concept underpinning the ethics of RCTs is equipoise. Here, we aimed to dissect the basic reasoning behind the concept of equipoise and we proposed a line of thinking delineating under which conditions it is ethical to design and execute an RCT. This might prevent a priori negative trials, reduce research waste and aid in the design of meaningful ones. It is these trials that will provide insight on how to safely and effectively assist subfertile couples.
Assuntos
Tomada de Decisão Clínica/ética , Ética em Pesquisa , Medicina Baseada em Evidências , Ensaios Clínicos Controlados Aleatórios como Assunto/ética , Medicina Reprodutiva/ética , Humanos , IncertezaRESUMO
The deep neural network, based on the backpropagation learning algorithm, has achieved tremendous success. However, the backpropagation algorithm is consistently considered biologically implausible. Many efforts have recently been made to address these biological implausibility issues, nevertheless, these methods are tailored to discrete neural network structures. Continuous neural networks are crucial for investigating novel neural network models with more biologically dynamic characteristics and for interpretability of large language models. The neural memory ordinary differential equation (nmODE) is a recently proposed continuous neural network model that exhibits several intriguing properties. In this study, we present a forward-learning algorithm, called nmForwardLA, for nmODE. This algorithm boasts lower computational dimensions and greater efficiency. Compared with the other learning algorithms, experimental results on MNIST, CIFAR10, and CIFAR100 demonstrate its potency.
Assuntos
Redes Neurais de Computação , Algoritmos , Humanos , Aprendizado Profundo , Aprendizado de MáquinaRESUMO
Generative models based on neural networks present a substantial challenge within deep learning. As it stands, such models are primarily limited to the domain of artificial neural networks. Spiking neural networks, as the third generation of neural networks, offer a closer approximation to brain-like processing due to their rich spatiotemporal dynamics. However, generative models based on spiking neural networks are not well studied. Particularly, previous works on generative adversarial networks based on spiking neural networks are conducted on simple datasets and do not perform well. In this work, we pioneer constructing a spiking generative adversarial network capable of handling complex images and having higher performance. Our first task is to identify the problems of out-of-domain inconsistency and temporal inconsistency inherent in spiking generative adversarial networks. We address these issues by incorporating the Earth-Mover distance and an attention-based weighted decoding method, significantly enhancing the performance of our algorithm across several datasets. Experimental results reveal that our approach outperforms existing methods on the MNIST, FashionMNIST, CIFAR10, and CelebA. In addition to our examination of static datasets, this study marks our inaugural investigation into event-based data, through which we achieved noteworthy results. Moreover, compared with hybrid spiking generative adversarial networks, where the discriminator is an artificial analog neural network, our methodology demonstrates closer alignment with the information processing patterns observed in the mouse. Our code can be found at https://github.com/Brain-Cog-Lab/sgad.
Assuntos
Algoritmos , Aprendizado Profundo , Redes Neurais de Computação , Humanos , Animais , Neurônios/fisiologia , Potenciais de Ação/fisiologia , Modelos Neurológicos , Atenção/fisiologia , Camundongos , Encéfalo/fisiologiaRESUMO
The environmental presence of pharmaceuticals, including the antidepressant fluoxetine, has become a subject of concern. Numerous studies have revealed effects of fluoxetine at environmental concentrations. Some of these studies have reported non-monotonic dose-response curves (NMDRs), leading to discussion because of the inconsistent detection of subtle effects and lack of mechanistic understanding. Nevertheless, investigating NMDRs in risk assessment is important, because neglecting them could underestimate potential risks of chemicals at low levels of exposure. Identification and quantification of NMDRs in risk assessment remains challenging, particularly given the prevalence of single outliers and the lack of sound statistical analyses. In response, the European Food Safety Authority (Beausoleil et al., 2016) presented a framework delineating six checkpoints for the evaluation of NMDR datasets, offering a systematic method for their assessment. The present study applies this framework to the case study of fluoxetine, aiming to assess the weight-of-evidence for the reported NMDR relationships. Through a systematic literature search, 53 datasets were selected for analysis against the six checkpoints. The results reveal that while a minority of these datasets meet all checkpoints, a significant proportion (27%) fulfilled at least five. Notably, many studies did not meet checkpoint 3, which requires NMDRs to be based on more than a single outlier. Overall, the current study points out a number of studies with considerable evidence supporting the presence of NMDRs for fluoxetine, while the majority of studies lacks strong evidence. The suggested framework proved useful for analysing NMDRs in ecotoxicological studies, but it is still imperative to develop further understanding of their biological plausibility.
Assuntos
Relação Dose-Resposta a Droga , Ecotoxicologia , Fluoxetina , Fluoxetina/toxicidade , Medição de Risco/métodos , Poluentes Ambientais/toxicidade , Humanos , AnimaisRESUMO
Isoflavones are naturally occurring plant compounds found in uniquely high amounts in soybeans and foods made from this legume. These soybean constituents have been proposed to exert several health benefits and as such they have been the subject of an enormous amount of research. This research includes randomized controlled trials (RCTs) and epidemiologic investigations. Although statistically significant associations between isoflavone intake and a wide range of health outcomes have been identified in cohorts involving low-isoflavone intake populations, we suggest that these associations are unlikely to have a causal basis because exposure is too low for isoflavones to exert physiologic effects. In cohorts involving predominantly non-Asian, non-vegetarian populations, the highest isoflavone intake category is typically ≤3 mg/d, an amount of isoflavones provided by â¼30 mL (2 tablespoons) of soymilk made from whole soybeans. In comparison, mean isoflavone intake in the upper intake categories in observational studies involving high-isoflavone intake populations is typically ≥50 mg/d. In RCTs, intervention doses of isoflavones typically range between 40 and 100 mg/d. Health professionals advising patients and clients about soy food and isoflavone intake need to be aware of the limitations of epidemiologic research involving low-isoflavone intake populations. Intake recommendations are best based on the results of RCTs using clinically relevant doses of isoflavones and epidemiologic studies involving populations for whom soy foods are a habitual part of the diet.
Assuntos
Glycine max , Isoflavonas , Humanos , Isoflavonas/farmacologia , Dieta , CetonasRESUMO
Objectives: Exposure to humidifier disinfectants has been linked to respiratory diseases, including interstitial lung disease, asthma, and pneumonia. Consequently, numerous toxicological studies have explored respiratory damage as both a necessary and sufficient condition for these diseases. We systematically reviewed and integrated evidence from toxicological studies by applying the evidence integration method established in previous research to confirm the biological plausibility of the association between exposure and disease. Methods: We conducted a literature search focusing on polyhexamethylene guanidine phosphate (PHMG) and chloromethylisothiazolinone/methylisothiazolinone (CMIT/MIT), the primary ingredients in humidifier disinfectants. We selected relevant studies based on their quality and the population, exposure, comparator, outcome (PECO) statements. These studies were categorized into 3 lines of evidence: hazard information, animal studies, and mechanistic studies. Based on a systematic review, we integrated the evidence to develop an aggregate exposure pathway-adverse outcome pathway (AEP-AOP) model for respiratory damage. The reliability and relevance of our findings were assessed by comparing them with the hypothesized pathogenic mechanisms of respiratory diseases. Results: The integration of each AEP-AOP component for PHMG and CMIT/MIT led to the development of an AEP-AOP model, wherein disinfectants released from humidifiers in aerosol or gaseous form reached target sites, causing respiratory damage through molecular initiating events and key events. The model demonstrated high reliability and relevance to the pathogenesis of respiratory diseases. Conclusion: The AEP-AOP model developed in this study provides strong evidence that exposure to humidifier disinfectants causes respiratory diseases. This model demonstrates the pathways leading to respiratory damage, a hallmark of these conditions.
RESUMO
A central goal in Cognitive Science is understanding the mechanisms that underlie cognition. Here, we contend that Cognitive Science, despite intense multidisciplinary efforts, has furnished surprisingly few mechanistic insights. We attribute this slow mechanistic progress to the fact that cognitive scientists insist on performing underdetermined exercises, deriving overparametrized mechanistic theories of complex behaviors and seeking validation of these theories to the elusive notions of optimality and biological plausibility. We propose that mechanistic progress in Cognitive Science will accelerate once cognitive scientists start focusing on simpler explananda that will enable them to chart an atlas of elementary cognitive operations. Looking forward, the next challenge for Cognitive Science will be to understand how these elementary cognitive processes are pieced together to explain complex behavior.
Assuntos
Cognição , Motivação , Humanos , Ciência CognitivaRESUMO
Emergence of deep neural networks (DNNs) has raised enormous attention towards artificial neural networks (ANNs) once again. They have become the state-of-the-art models and have won different machine learning challenges. Although these networks are inspired by the brain, they lack biological plausibility, and they have structural differences compared to the brain. Spiking neural networks (SNNs) have been around for a long time, and they have been investigated to understand the dynamics of the brain. However, their application in real-world and complicated machine learning tasks were limited. Recently, they have shown great potential in solving such tasks. Due to their energy efficiency and temporal dynamics there are many promises in their future development. In this work, we reviewed the structures and performances of SNNs on image classification tasks. The comparisons illustrate that these networks show great capabilities for more complicated problems. Furthermore, the simple learning rules developed for SNNs, such as STDP and R-STDP, can be a potential alternative to replace the backpropagation algorithm used in DNNs.â¢Different building blocks of spiking neural networks are explained in this work.â¢Developed models for SNNs are introduced based on their characteristics and building blocks.
RESUMO
In deep neural networks, representational learning in the middle layer is essential for achieving efficient learning. However, the currently prevailing backpropagation learning rules (BP) are not necessarily biologically plausible and cannot be implemented in the brain in their current form. Therefore, to elucidate the learning rules used by the brain, it is critical to establish biologically plausible learning rules for practical memory tasks. For example, learning rules that result in a learning performance worse than that of animals observed in experimental studies may not be computations used in real brains and should be ruled out. Using numerical simulations, we developed biologically plausible learning rules to solve a task that replicates a laboratory experiment where mice learned to predict the correct reward amount. Although the extreme learning machine (ELM) and weight perturbation (WP) learning rules performed worse than the mice, the feedback alignment (FA) rule achieved a performance equal to that of BP. To obtain a more biologically plausible model, we developed a variant of FA, FA_Ex-100%, which implements direct dopamine inputs that provide error signals locally in the layer of focus, as found in the mouse entorhinal cortex. The performance of FA_Ex-100% was comparable to that of conventional BP. Finally, we tested whether FA_Ex-100% was robust against rule perturbations and biologically inevitable noise. FA_Ex-100% worked even when subjected to perturbations, presumably because it could calibrate the correct prediction error (e.g., dopaminergic signals) in the next step as a teaching signal if the perturbation created a deviation. These results suggest that simplified and biologically plausible learning rules, such as FA_Ex-100%, can robustly facilitate deep supervised learning when the error signal, possibly conveyed by dopaminergic neurons, is accurate.
RESUMO
This article presents a comprehensive analysis of spiking neural networks (SNNs) and their mathematical models for simulating the behavior of neurons through the generation of spikes. The study explores various models, including LIF and NLIF, for constructing SNNs and investigates their potential applications in different domains. However, implementation poses several challenges, including identifying the most appropriate model for classification tasks that demand high accuracy and low-performance loss. To address this issue, this research study compares the performance, behavior, and spike generation of multiple SNN models using consistent inputs and neurons. The findings of the study provide valuable insights into the benefits and challenges of SNNs and their models, emphasizing the significance of comparing multiple models to identify the most effective one. Moreover, the study quantifies the number of spiking operations required by each model to process the same inputs and produce equivalent outputs, enabling a thorough assessment of computational efficiency. The findings provide valuable insights into the benefits and limitations of SNNs and their models. The research underscores the significance of comparing different models to make informed decisions in practical applications. Additionally, the results reveal essential variations in biological plausibility and computational efficiency among the models, further emphasizing the importance of selecting the most suitable model for a given task. Overall, this study contributes to a deeper understanding of SNNs and offers practical guidelines for using their potential in real-world scenarios.
RESUMO
The core of bodily self-consciousness involves perceiving ownership of one's body. A central question is how body illusions like the rubber hand illusion (RHI) occur. Existing theoretical models still lack satisfying computational explanations from connectionist perspectives, especially for how the brain encodes body perception and generates illusions from neuronal interactions. Moreover, the integration of disability experiments is also neglected. Here, we integrate biological findings of bodily self-consciousness to propose a brain-inspired bodily self-perception model by which perceptions of bodily self are autonomously constructed without any supervision signals. We successfully validated the model with six RHI experiments and a disability experiment on an iCub humanoid robot and simulated environments. The results show that our model can not only well-replicate the behavioral and neural data of monkeys in biological experiments but also reasonably explain the causes and results of RHI at the neuronal level, thus contributing to the revelation of mechanisms underlying RHI.
RESUMO
BACKGROUND: "Biological plausibility" is a concept frequently referred to in environmental and public health when researchers are evaluating how confident they are in the results and inferences of a study or evidence review. Biological plausibility is not, however, a domain of one of the most widely-used approaches for assessing the certainty of evidence (CoE) which underpins the findings of a systematic review, the Grading of Recommendations Assessment, Development and Evaluation (GRADE) CoE Framework. Whether the omission of biological plausibility is a potential limitation of the GRADE CoE Framework is a topic that is regularly discussed, especially in the context of environmental health systematic reviews. OBJECTIVES: We analyse how the concept of "biological plausibility", as applied in the context of assessing certainty of the evidence that supports the findings of a systematic review, is accommodated under the processes of systematic review and the existing GRADE domains. RESULTS AND DISCUSSION: We argue that "biological plausibility" is a concept which primarily comes into play when direct evidence about the effects of an exposure on a population of concern (usually humans) is absent, at high risk of bias, is inconsistent, or limited in other ways. In such circumstances, researchers look toward evidence from other study designs in order to draw conclusions. In this respect, we can consider experimental animal and in vitro evidence as "surrogates" for the target populations, exposures, comparators and outcomes of actual interest. Through discussion of 10 examples of experimental surrogates, we propose that the concept of biological plausibility consists of two principal aspects: a "generalisability aspect" and a "mechanistic aspect". The "generalisability aspect" concerns the validity of inferences from experimental models to human scenarios, and asks the same question as does the assessment of external validity or indirectness in systematic reviews. The "mechanistic aspect" concerns certainty in knowledge of biological mechanisms and would inform judgements of indirectness under GRADE, and thus the overall CoE. While both aspects are accommodated under the indirectness domain of the GRADE CoE Framework, further research is needed to determine how to use knowledge of biological mechanisms in the assessment of indirectness of the evidence in systematic reviews.
Assuntos
Saúde Ambiental , Abordagem GRADE , Animais , Viés , Saúde Pública , Revisões Sistemáticas como AssuntoRESUMO
Neurons responding sensitively to motions in several rather than all directions have been identified in many sensory systems. Although this directional preference has been demonstrated by previous studies to exist in the isthmi pars magnocellularis (Imc) of pigeon (Columba livia), which plays a key role in the midbrain saliency computing network, the dynamic response characteristics and the physiological basis underlying this phenomenon are unclear. Herein, dots moving in 16 directions and a biologically plausible computational model were used. We found that pigeon Imc's significant responses for objects moving in preferred directions benefit the long response duration and high instantaneous firing rate. Furthermore, the receptive field structures predicted by a computational model, which captures the actual directional tuning curves, agree with the real data collected from population Imc units. These results suggested that directional preference in Imc may be internally prebuilt by elongating the vertical axis of the receptive field, making predators attack from the dorsal-ventral direction and conspecifics flying away in the ventral-dorsal direction, more salient for avians, which is of great ecological and physiological significance for survival.
RESUMO
The number of gene therapies in clinical trials and moving toward licensure is increasing. Most gene therapies are designed to achieve long-term effects, but at licensure the data to support claims of long-term durability are often limited, as long-term monitoring studies are often part of post-approval commitments by companies. Health technology assessors must therefore assess the potential for the long-term durability of a product and the potential cost-effectiveness based on the data available. The authors explored the benefit of strengthening the ability to infer durability of effect using analogue category data. Different analogue categories were assessed for the potential to substantiate claims of sustainability of effect for gene therapies by leveraging biological plausibility arguments. The authors propose a pathway for identifying potential analogues. Such a pathway should help establish plausible or theoretical long-term outcomes that can be considered in value assessments of gene therapies.
Many diseases, affecting all parts of the body, can be treated with gene therapy. Gene therapies make changes to a person's genes by either replacing or inactivating a gene that is causing disease or by adding new genes that can fight diseases such as cancers. These therapies have the potential to cure patients of the disease for their lifetime. When decisions are being made over whether gene therapies are safe and work well in patients, it can be difficult to understand if they will maintain their benefits to a person over a lifetime, as they have only been studied in clinical trials for a much shorter amount of time. In this paper, the authors explore whether information around the benefits of therapies that work in a similar way, or target similar diseases, can be used to strengthen an understanding of how well newer therapies work over a long period of time. The authors propose a pathway that can be followed to identify the suitability of a comparison between different therapies and how the evidence of benefit over time can be interpreted. This information will be useful for developers of gene therapies who are trying to generate evidence of long-term benefit to patients, as well as for decision makers who need to understand how well these gene therapies will work over a patient's lifetime.
Assuntos
Análise Custo-BenefícioRESUMO
BACKGROUND: "Biological plausibility" is a concept frequently referred to in environmental and public health when researchers are evaluating how confident they are in the results and inferences of a study or evidence review. Biological plausibility is not, however, a domain of one of the most widely used approaches for assessing the certainty of evidence (CoE) which underpins the findings of a systematic review, the Grading of Recommendations, Assessment, Development and Evaluation (GRADE) CoE Framework. OBJECTIVES: Whether the omission of biological plausibility is a potential limitation of the GRADE CoE Framework is a topic that is regularly discussed, especially in the context of environmental health systematic reviews. STUDY DESIGN AND SETTING: We analyze how the concept of "biological plausibility," as applied in the context of assessing certainty of the evidence that supports the findings of a systematic review, is accommodated under the processes of systematic review and the existing GRADE domains. RESULTS: We argue that "biological plausibility" is a concept which primarily comes into play when direct evidence about the effects of an exposure on a population of concern (usually humans) is absent, at high risk of bias, inconsistent, or limited in other ways. In such circumstances, researchers look toward evidence from other study designs to draw conclusions. In this respect, we can consider experimental animal and in vitro evidence as "surrogates" for the target populations, exposures, comparators, and outcomes of actual interest. Through discussion of 10 examples of experimental surrogates, we propose that the concept of biological plausibility consists of two principal aspects: a "generalizability aspect" and a "mechanistic aspect." CONCLUSIONS: The "generalizability aspect" concerns the validity of inferences from experimental models to human scenarios, and asks the same question as does the assessment of external validity or indirectness in systematic reviews. The "mechanistic aspect" concerns certainty in knowledge of biological mechanisms and would inform judgments of indirectness under GRADE, and thus the overall CoE. Although both aspects are accommodated under the indirectness domain of the GRADE CoE Framework, further research is needed to determine how to use knowledge of biological mechanisms in the assessment of indirectness of the evidence in systematic reviews.
Assuntos
Saúde Ambiental , Abordagem GRADE , Animais , Viés , Humanos , Projetos de Pesquisa , Revisões Sistemáticas como AssuntoRESUMO
Notions of mechanism, emergence, reduction and explanation are all tied to levels of analysis. I cover the relationship between lower and higher levels, suggest a level of mechanism approach for neuroscience in which the components of a mechanism can themselves be further decomposed and argue that scientists' goals are best realized by focusing on pragmatic concerns rather than on metaphysical claims about what is 'real'. Inexplicably, neuroscientists are enchanted by both reduction and emergence. A fascination with reduction is misplaced given that theory is neither sufficiently developed nor formal to allow it, whereas metaphysical claims of emergence bring physicalism into question. Moreover, neuroscience's existence as a discipline is owed to higher-level concepts that prove useful in practice. Claims of biological plausibility are shown to be incoherent from a level of mechanism view and more generally are vacuous. Instead, the relevant findings to address should be specified so that model selection procedures can adjudicate between competing accounts. Model selection can help reduce theoretical confusions and direct empirical investigations. Although measures themselves, such as behaviour, blood-oxygen-level-dependent (BOLD) and single-unit recordings, are not levels of analysis, like levels, no measure is fundamental and understanding how measures relate can hasten scientific progress. This article is part of the theme issue 'Key relationships between non-invasive functional neuroimaging and the underlying neuronal activity'.
Assuntos
Encéfalo/fisiologia , Neurônios/fisiologia , HumanosRESUMO
Sedentary behavior, particularly sitting, is ubiquitous in many contemporary societies. This is a major societal concern considering the evidence for a strong association between sitting behavior and cardiovascular disease morbidity and mortality. Unsurprisingly, leading public health agencies have begun to advocate "reduction" in sitting behavior. Though, the guidelines are typically vague and non-specific. The lack of specific guidelines for prolonged sitting is attributable to the absence of available evidence to facilitate guideline development. To inform policy, well-designed randomized controlled trials are required to test the efficacy of specific and translatable sitting reduction strategies. To guide the design of randomized controlled trials, this review postulates that several gaps in the literature first need to be filled. Following a general discussion of the importance of sitting behavior to contemporary societies, each of the following are discussed: (i) acute sitting exposure and systems physiology; (ii) recommendations for a systems physiology toolbox; (iii) study design considerations for acute sitting exposure; and (iv) translation of sitting-focused research.
RESUMO
Modern Machine learning techniques take advantage of the exponentially rising calculation power in new generation processor units. Thus, the number of parameters which are trained to solve complex tasks was highly increased over the last decades. However, still the networks fail - in contrast to our brain - to develop general intelligence in the sense of being able to solve several complex tasks with only one network architecture. This could be the case because the brain is not a randomly initialized neural network, which has to be trained from scratch by simply investing a lot of calculation power, but has from birth some fixed hierarchical structure. To make progress in decoding the structural basis of biological neural networks we here chose a bottom-up approach, where we evolutionarily trained small neural networks in performing a maze task. This simple maze task requires dynamic decision making with delayed rewards. We were able to show that during the evolutionary optimization random severance of connections leads to better generalization performance of the networks compared to fully connected networks. We conclude that sparsity is a central property of neural networks and should be considered for modern Machine learning approaches.