How humans learn and represent networks.
Ontology highlight
ABSTRACT:
SUBMITTER: Lynn CW
PROVIDER: S-EPMC7703562 | biostudies-literature | 2020 Nov
REPOSITORIES: biostudies-literature
Publications
How humans learn and represent networks.
Lynn Christopher W CW Bassett Danielle S DS
Proceedings of the National Academy of Sciences of the United States of America 20201101 47
Similar Datasets
Project description:Popular computational models of decision-making make specific assumptions about learning processes that may cause them to underfit observed behaviours. Here we suggest an alternative method using recurrent neural networks (RNNs) to generate a flexible family of models that have sufficient capacity to represent the complex learning and decision- making strategies used by humans. In this approach, an RNN is trained to predict the next action that a subject will take in a decision-making task and, in this way, learns to imitate the processes underlying subjects' choices and their learning abilities. We demonstrate the benefits of this approach using a new dataset drawn from patients with either unipolar (n = 34) or bipolar (n = 33) depression and matched healthy controls (n = 34) making decisions on a two-armed bandit task. The results indicate that this new approach is better than baseline reinforcement-learning methods in terms of overall performance and its capacity to predict subjects' choices. We show that the model can be interpreted using off-policy simulations and thereby provides a novel clustering of subjects' learning processes-something that often eludes traditional approaches to modelling and behavioural analysis.
| S-EPMC6588260 | biostudies-literature
Project description:Practically all animals are affected by humans, especially in urban areas. Although most species respond negatively to urbanization, some thrive in human-dominated settings. A central question in urban ecology is why some species adapt well to the presence of humans and others do not. We show that Northern Mockingbirds (Mimus polyglottos) nesting on the campus of a large university rapidly learn to assess the level of threat posed by different humans, and to respond accordingly. In a controlled experiment, we found that as the same human approached and threatened a nest on 4 successive days, mockingbirds flushed from their nest at increasingly greater distances from that human. A different human approaching and threatening the nest identically on the fifth day elicited the same response as the first human on the first day. Likewise, alarm calls and attack flights increased from days 1-4 with the first human, and decreased on day 5 with the second human. These results demonstrate a remarkable ability of a passerine bird to distinguish one human from thousands of others. Also, mockingbirds learned to identify individual humans extraordinarily quickly: after only 2 30-s exposures of the human at the nest. More generally, the varying responses of mockingbirds to intruders suggests behavioral flexibility and a keen awareness of different levels of threat posed by individuals of another species: traits that may predispose mockingbirds and other species of urban wildlife to successful exploitation of human-dominated environments.
| S-EPMC2690012 | biostudies-literature
Project description:Lamin proteins are type V intermediate filament proteins (IFs) located inside the cell nucleus. They are evolutionarily conserved and have similar domain organization and properties to cytoplasmic IFs. Lamins provide a skeletal network for chromatin, the nuclear envelope, nuclear pore complexes and the entire nucleus. They are also responsible for proper connections between the karyoskeleton and structural elements in the cytoplasm: actin and the microtubule and cytoplasmic IF networks. Lamins affect transcription and splicing either directly or indirectly. Translocation of active genes into the close proximity of nuclear lamina is thought to result in their transcriptional silencing. Mutations in genes coding for lamins and interacting proteins in humans result in various genetic disorders, called laminopathies. Human genes coding for A-type lamin (LMNA) are the most frequently mutated. The resulting phenotypes include muscle, cardiac, neuronal, lipodystrophic and metabolic pathologies, early aging phenotypes, and combined complex phenotypes. The Drosophila melanogaster genome codes for lamin B-type (lamin Dm), lamin A-type (lamin C), and for LEM-domain proteins, BAF, LINC-complex proteins and all typical nuclear proteins. The fruit fly system is simpler than the vertebrate one since in flies there is only single lamin B-type and single lamin A-type protein, as opposed to the complex system of B- and A-type lamins in Danio, Xenopus and Mus musculus. This offers a unique opportunity to study laminopathies. Applying genetic tools based on Gal4 and in vitro nuclear assembly system to the fruit fly model may successfully advance knowledge of laminopathies. Here, we review studies of the laminopathies in the fly model system.
| S-EPMC6034310 | biostudies-literature
Project description:Animals can acquire new behavior through both individual and social learning. Several studies have investigated horses' ability to utilize inter-species (human demonstrator) social learning with conflicting results. In this study, we repeat a previous study, which found that horses had the ability to learn from observing humans performing an instrumental task, but we include a control for stimulus enhancement. One human demonstrator and thirty horses were included, and the horses were randomly assigned to one of three treatments: (A) full human demonstration, (B) partial human demonstration, and (C) no human demonstration. The task was for the horses to touch an object situated 1 m away from a feed box, to open this feed box, and thereby obtain a food reward. The success of each horse, the behavior directed towards the apparatus and the human, and behaviors indicative of frustration were observed. The results showed that horses observing a full and partial human demonstration were not more successful in solving the instrumental task than horses not observing any demonstration. Horses that did not solve the task expressed more box- and human-oriented behavior compared to successful horses, which may be an indication of motivation to solve the task and/or frustration from being unable to solve the task.
| S-EPMC7070367 | biostudies-literature
Project description:Machine learning is playing an increasing role in the physical sciences and significant progress has been made towards embedding domain knowledge into models. Less explored is its use to discover interpretable physical laws from data. We propose parsimonious neural networks (PNNs) that combine neural networks with evolutionary optimization to find models that balance accuracy with parsimony. The power and versatility of the approach is demonstrated by developing models for classical mechanics and to predict the melting temperature of materials from fundamental properties. In the first example, the resulting PNNs are easily interpretable as Newton's second law, expressed as a non-trivial time integrator that exhibits time-reversibility and conserves energy, where the parsimony is critical to extract underlying symmetries from the data. In the second case, the PNNs not only find the celebrated Lindemann melting law, but also new relationships that outperform it in the pareto sense of parsimony vs. accuracy.
| S-EPMC8211802 | biostudies-literature
Project description:With the goal of understanding behavioral mechanisms of generalization, we analyzed the ability of neural networks to generalize across context. We modeled a behavioral task where the correct responses to a set of specific sensory stimuli varied systematically across different contexts. The correct response depended on the stimulus (A,B,C,D) and context quadrant (1,2,3,4). The possible 16 stimulus-context combinations were associated with one of two responses (X,Y), one of which was correct for half of the combinations. The correct responses varied symmetrically across contexts. This allowed responses to previously unseen stimuli (probe stimuli) to be generalized from stimuli that had been presented previously. By testing the simulation on two or more stimuli that the network had never seen in a particular context, we could test whether the correct response on the novel stimuli could be generated based on knowledge of the correct responses in other contexts. We tested this generalization capability with a Deep Belief Network (DBN), Multi-Layer Perceptron (MLP) network, and the combination of a DBN with a linear perceptron (LP). Overall, the combination of the DBN and LP had the highest success rate for generalization.
| S-EPMC3966868 | biostudies-literature
Project description:In a range of settings, human operators make decisions with the assistance of automation, the reliability of which can vary depending upon context. Currently, the processes by which humans track the level of reliability of automation are unclear. In the current study, we test cognitive models of learning that could potentially explain how humans track automation reliability. We fitted several alternative cognitive models to a series of participants' judgements of automation reliability observed in a maritime classification task in which participants were provided with automated advice. We examined three experiments including eight between-subjects conditions and 240 participants in total. Our results favoured a two-kernel delta-rule model of learning, which specifies that humans learn by prediction error, and respond according to a learning rate that is sensitive to environmental volatility. However, we found substantial heterogeneity in learning processes across participants. These outcomes speak to the learning processes underlying how humans estimate automation reliability and thus have implications for practice.
| S-EPMC10869332 | biostudies-literature
Project description:Most existing protected area networks are biased to protect charismatic species or landscapes. We hypothesized that conservation networks designed to include unseen biodiversity-species rich groups that consist of inconspicuous taxa, or groups affected by knowledge gaps-are more efficient than networks that ignore these groups. To test this hypothesis, we generated species distribution models for 3006 arthropod species to determine which were represented in three networks of different sizes and biogeographic origin. We assessed the efficiency of each network using spatial prioritization to measure its completeness, the increment needed to achieve conservation targets, and its specificity, the extent to which proposed priority areas to maximize unseen biodiversity overlap with existing networks. We found that the representativeness of unseen biodiversity in the studied protected areas, or extrinsic representativeness, is low, with ~ 40% of the analyzed unseen biodiversity species being unprotected. We also found that existing networks should be expanded ~ 26% to 46% of their current area to complete targets, and that existing networks do not efficiently conserve the unseen biodiversity given their low specificity (as low as 8.8%) unseen biodiversity. We conclude that information on unseen biodiversity must be included in systematic conservation planning approaches to design more efficient and ecologically representative protected areas.
| S-EPMC8192537 | biostudies-literature
Project description:Our environment contains an abundance of objects which humans interact with daily, gathering visual information using sequences of eye-movements to choose which object is best-suited for a particular task. This process is not trivial, and requires a complex strategy where task affordance defines the search strategy, and the estimated precision of the visual information gathered from each object may be used to track perceptual confidence for object selection. This study addresses the fundamental problem of how such visual information is metacognitively represented and used for subsequent behaviour, and reveals a complex interplay between task affordance, visual information gathering, and metacogntive decision making. People fixate higher-utility objects, and most importantly retain metaknowledge about how much information they have gathered about these objects, which is used to guide perceptual report choices. These findings suggest that such metacognitive knowledge is important in situations where decisions are based on information acquired in a temporal sequence.
| S-EPMC8844410 | biostudies-literature
Project description:We demonstrate that neural networks that process noisy data can learn to exploit, when available, access to auxiliary noise that is correlated with the noise on the data. In effect, the network learns to use the correlated auxiliary noise as an approximate key to decipher its noisy input data. An example of naturally occurring correlated auxiliary noise is the noise due to decoherence. Our results could, therefore, also be of interest, for example, for machine-learned quantum error correction.
| S-EPMC8566565 | biostudies-literature