The pigeon as a machine: Complex category structures can be acquired by a simple associative model
Ontology highlight
ABSTRACT:
SUBMITTER: Turner B
PROVIDER: S-EPMC10579431 | biostudies-literature | 2023 Sep
REPOSITORIES: biostudies-literature
Similar Datasets
Project description:Adaptively and flexibly modifying one's behavior depending on the current demands of the situation is a hallmark of executive function. Here, we examined whether pigeons could flexibly shift their attention from one set of features that were relevant in one categorization task to another set of features that were relevant in a second categorization task. Critically, members of both sets of features were available on every training trial, thereby requiring that attention be adaptively deployed on a trial-by-trial basis based on contextual information. The pigeons not only learned to correctly categorize the stimuli but, as training progressed, they concentrated their pecks to the training stimuli (a proxy measure for attention) on those features that were relevant in a specific context. The pigeons selectively tracked the features that were relevant in Context 1-but were irrelevant in Context 2-and they selectively tracked the features that were relevant in Context 2-but were irrelevant in Context 1. This adept feature tracking requires disengaging attention from a previously relevant feature and shifting attention to a previously ignored feature on a trial-by-trial basis. Pigeons' adaptive and flexible performance provides strong empirical support for the involvement of focusing and shifting attention under exceptionally challenging training conditions. (PsycInfo Database Record (c) 2021 APA, all rights reserved).
| S-EPMC8514057 | biostudies-literature
Project description:Attention to relevant stimulus features in a categorization task helps to optimize performance. However, the relationship between attention and categorization is not fully understood. For example, even when human adults and young children exhibit comparable categorization behavior, adults tend to attend selectively during learning, whereas young children tend to attend diffusely (Deng & Sloutsky, 2016). Here, we used a comparative approach to investigate the link between attention and categorization in two different species. Given the noteworthy categorization ability of avian species, we compared the attentional profiles of pigeons and human adults. We gave human adults (Experiment 1) and pigeons (Experiment 2) a categorization task that could be learned on the basis of either one deterministic feature (encouraging selective attention) or multiple probabilistic features (encouraging distributed attention). Both humans and pigeons relied on the deterministic feature to categorize the stimuli, albeit humans did so to a much greater degree. Furthermore, computational modeling revealed that most of the adults exhibited maximal selectivity, whereas pigeons tended to distribute their attention among several features. Our findings indicate that human adults focus their attention on deterministic information and filter less predictive information, but pigeons do not. Implications for the underlying brain mechanisms of attention and categorization are discussed.
| S-EPMC7494638 | biostudies-literature
Project description:In a seminal study, Shepard, Hovland, and Jenkins (1961; henceforth SHJ) assessed potential mechanisms involved in categorization learning. To do so, they sequentially trained human participants with 6 different visual categorization tasks that varied in structural complexity. Humans' exceptionally strong performance on 1 of these tasks (Type 2, organized around exclusive-or relations) could not be solely explained by structural complexity, and has since been considered the hallmark of rule-use in these tasks. In the present project, we concurrently trained pigeons on all 6 SHJ tasks. Our results revealed that the structural complexity of the tasks was highly correlated with group-level performance. Nevertheless, we observed notable individual differences in performance. Two extensions of a prominent categorization model, ALCOVE (Kruschke, 1992), suggested that disparities in the discriminability of the dimensions used to construct the experimental stimuli could account for these differences. Overall, our pigeons' generally weak performance on the Type 2 task provides no evidence of rule-use on the SHJ tasks. Pigeons thus join monkeys in the contingent of species that solve these categorization tasks solely on the basis of the physical properties of the training stimuli. (PsycINFO Database Record (c) 2019 APA, all rights reserved).
| S-EPMC6730555 | biostudies-literature
Project description:A wealth of evidence indicates that humans can engage two types of mechanisms to solve category-learning tasks: declarative mechanisms, which involve forming and testing verbalizable decision rules, and associative mechanisms, which involve gradually linking stimuli to appropriate behavioral responses.1,2,3 In contrast to declarative mechanisms, associative mechanisms have received surprisingly little attention in the broader category-learning literature. Although various forms of associatively driven artificial intelligence (AI) have matched-and even surpassed-humans' performance on several challenging problems,3,4,5,6 associative learning is routinely dismissed as being too simple to power the impressive cognitive achievements of both humans and non-human species.6,7,8,9 Here, we attempt to resolve this paradox by demonstrating that pigeons-which appear to rely solely on associative learning mechanisms in several tasks that promote declarative rule use by humans3,10,11,12-succeed at learning a novel, highly demanding category structure that ought to hinder declarative rule use: the sectioned-rings task. Our findings highlight the power and flexibility that associative mechanisms afford in the realm of category learning.
| S-EPMC10050111 | biostudies-literature
Project description:Simulation techniques based on accurate and efficient representations of potential energy surfaces are urgently needed for the understanding of complex systems such as solid-liquid interfaces. Here we present a machine learning framework that enables the efficient development and validation of models for complex aqueous systems. Instead of trying to deliver a globally optimal machine learning potential, we propose to develop models applicable to specific thermodynamic state points in a simple and user-friendly process. After an initial ab initio simulation, a machine learning potential is constructed with minimum human effort through a data-driven active learning protocol. Such models can afterward be applied in exhaustive simulations to provide reliable answers for the scientific question at hand or to systematically explore the thermal performance of ab initio methods. We showcase this methodology on a diverse set of aqueous systems comprising bulk water with different ions in solution, water on a titanium dioxide surface, and water confined in nanotubes and between molybdenum disulfide sheets. Highlighting the accuracy of our approach with respect to the underlying ab initio reference, the resulting models are evaluated in detail with an automated validation protocol that includes structural and dynamical properties and the precision of the force prediction of the models. Finally, we demonstrate the capabilities of our approach for the description of water on the rutile titanium dioxide (110) surface to analyze the structure and mobility of water on this surface. Such machine learning models provide a straightforward and uncomplicated but accurate extension of simulation time and length scales for complex systems.
| S-EPMC8463804 | biostudies-literature
Project description:Humans appear to represent many forms of knowledge in associative networks whose nodes are multiply connected, including sensory, spatial, and semantic. Recent work has shown that explicitly augmenting artificial agents with such graph-structured representations endows them with more human-like capabilities of compositionality and transfer learning. An open question is how humans acquire these representations. Previously, it has been shown that humans can learn to navigate graph-structured conceptual spaces on the basis of direct experience with trajectories that intentionally draw the network contours (Schapiro, Kustner, & Turk-Browne, 2012; Schapiro, Turk-Browne, Botvinick, & Norman, 2016), or through direct experience with rewards that covary with the underlying associative distance (Wu, Schulz, Speekenbrink, Nelson, & Meder, 2018). Here, we provide initial evidence that this capability is more general, extending to learning to reason about shortest-path distances across a graph structure acquired across disjoint experiences with randomized edges of the graph - a form of latent learning. In other words, we show that humans can infer graph structures, assembling them from disordered experiences. We further show that the degree to which individuals learn to reason correctly and with reference to the structure of the graph corresponds to their propensity, in a separate task, to use model-based reinforcement learning to achieve rewards. This connection suggests that the correct acquisition of graph-structured relationships is a central ability underlying forward planning and reasoning, and may be a core computation across the many domains in which graph-based reasoning is advantageous.
| S-EPMC9201735 | biostudies-literature
Project description:Nanoparticles with "sticky patches" have long been proposed as building blocks for the self-assembly of complex structures. The synthetic realizability of such patchy particles, however, greatly lags behind predictions of patterns they could form. Using computer simulations, we show that structures of the same genre can be obtained from a solution of simple isotropic spheres, with control only over their sizes and a small number of binding affinities. In a first step, finite clusters of well-defined structure and composition emerge from natural dynamics with high yield. In effect a kind of patchy particle, these clusters can further assemble into a variety of complex superstructures, including filamentous networks, ordered sheets, and highly porous crystals.
| S-EPMC4071109 | biostudies-literature
Project description:Real-world networks such as the Internet and WWW have many common traits. Until now, hundreds of models were proposed to characterize these traits for understanding the networks. Because different models used very different mechanisms, it is widely believed that these traits origin from different causes. However, we find that a simple model based on optimisation can produce many traits, including scale-free, small-world, ultra small-world, Delta-distribution, compact, fractal, regular and random networks. Moreover, by revising the proposed model, the community-structure networks are generated. By this model and the revised versions, the complicated relationships of complex networks are illustrated. The model brings a new universal perspective to the understanding of complex networks and provide a universal method to model complex networks from the viewpoint of optimisation.
| S-EPMC4145283 | biostudies-literature
Project description:Linear drug toxicity models like therapeutic index (TI), physicochemical rules (rule of five, 3/75), ligand efficiency indices (LEI), ideal pharmacokinetic (PK) and pharmacodynamic (PD) profiles are widely used in drug discovery and development. In spite of this, predicting drug toxicity at various stages remains challenging and the overall productivity (<20%) and ultimate benefit to the patients remain low. A simple drug toxicity model, "Drug Toxicity Index" (DTI), is developed here using 711 oral drugs. DTI redefines drug toxicity as scaled biphasic and exponential functions of PD, PK and physicochemical parameters. PD, PK and physicochemical toxicity contributions were estimated from the on and off target IC50, maximum unbound plasma drug concentration (free C max), and log D values, respectively. These contributions are then scaled by molar dose and oral bioavailability and the logarithm of the sum of scaled contributions is DTI. Drugs with DTI above the WHO ATC drug category specific average values consistently have toxic profiles, while drugs with DTI below this average are relatively safe. DTI performs better than standard rules for lead optimization, LEI and exposure based TIs in identifying safe and toxic drugs. DTI classifies 392 drugs reported in the US-FDA's Liver Toxicity Knowledge Base (LTKB) with an AUC for ROC curves of 0.91-0.64 for different WHO ATC categories. DTI has been used to predict network meta-analysis results on relative toxicity within/across eight different therapeutic areas. It is useful in understanding PD, PK and physicochemical toxicity contributions and identifying potentially toxic drugs and the toxicity of recently approved drugs. Decision trees are proposed for applying the DTI concept in preclinical drug discovery and clinical trial settings. DTI can potentially reduce failure in drug discovery and might be useful in therapeutic drug monitoring and in xenobiotic and environmental toxicity studies.
| S-EPMC6417485 | biostudies-literature
Project description:Insect wings are typically supported by thickened struts called veins. These veins form diverse geometric patterns across insects. For many insect species, even the left and right wings from the same individual have veins with unique topological arrangements, and little is known about how these patterns form. We present a large-scale quantitative study of the fingerprint-like "secondary veins." We compile a dataset of wings from 232 species and 17 families from the order Odonata (dragonflies and damselflies), a group with particularly elaborate vein patterns. We characterize the geometric arrangements of veins and develop a simple model of secondary vein patterning. We show that our model is capable of recapitulating the vein geometries of species from other, distantly related winged insect clades.
| S-EPMC6176563 | biostudies-literature