Dataset Information

A model of collective behavior based purely on vision.

ABSTRACT: Classical models of collective behavior often take a "bird's-eye perspective," assuming that individuals have access to social information that is not directly available (e.g., the behavior of individuals outside of their field of view). Despite the explanatory success of those models, it is now thought that a better understanding needs to incorporate the perception of the individual, i.e., how internal and external information are acquired and processed. In particular, vision has appeared to be a central feature to gather external information and influence the collective organization of the group. Here, we show that a vision-based model of collective behavior is sufficient to generate organized collective behavior in the absence of spatial representation and collision. Our work suggests a different approach for the development of purely vision-based autonomous swarm robotic systems and formulates a mathematical framework for exploration of perception-based interactions and how they differ from physical ones.

SUBMITTER: Bastien R

PROVIDER: S-EPMC7002123 | biostudies-literature | 2020 Feb

REPOSITORIES: biostudies-literature

ACCESS DATA

Publications

A model of collective behavior based purely on vision.

Bastien Renaud R Romanczuk Pawel P

Science advances 20200205 6

Classical models of collective behavior often take a "bird's-eye perspective," assuming that individuals have access to social information that is not directly available (e.g., the behavior of individuals outside of their field of view). Despite the explanatory success of those models, it is now thought that a better understanding needs to incorporate the perception of the individual, i.e., how internal and external information are acquired and processed. In particular, vision has appeared to be ...[more]

PMID: 32076645

Similar Datasets

Project description:Many species of fish rely on their visual systems to interact with conspecifics and these interactions can lead to collective behavior. Individual-based models have been used to predict collective interactions; however, these models generally make simplistic assumptions about the sensory systems that are applied without proper empirical testing to different species. This could limit our ability to predict (and test empirically) collective behavior in species with very different sensory requirements. In this study, we characterized components of the visual system in two species of cyprinid fish known to engage in visually dependent collective interactions (zebrafish Danio rerio and golden shiner Notemigonus crysoleucas) and derived quantitative predictions about the positioning of individuals within schools. We found that both species had relatively narrow binocular and blind fields and wide visual coverage. However, golden shiners had more visual coverage in the vertical plane (binocular field extending behind the head) and higher visual acuity than zebrafish. The centers of acute vision (areae) of both species projected in the fronto-dorsal region of the visual field, but those of the zebrafish projected more dorsally than those of the golden shiner. Based on this visual sensory information, we predicted that: (a) predator detection time could be increased by >1,000% in zebrafish and >100% in golden shiners with an increase in nearest neighbor distance, (b) zebrafish schools would have a higher roughness value (surface area/volume ratio) than those of golden shiners, (c) and that nearest neighbor distance would vary from 8 to 20 cm to visually resolve conspecific striping patterns in both species. Overall, considering between-species differences in the sensory system of species exhibiting collective behavior could change the predictions about the positioning of individuals in the group as well as the shape of the school, which can have implications for group cohesion. We suggest that more effort should be invested in assessing the role of the sensory system in shaping local interactions driving collective behavior.

Project description:What motivates an action in the absence of a definite reward? Taking the case of visuomotor control, we consider a minimal control problem that is how select the next saccade, in a sequence of discrete eye movements, when the final objective is to better interpret the current visual scene. The visual scene is modeled here as a partially-observed environment, with a generative model explaining how the visual data is shaped by action. This allows to interpret different action selection metrics proposed in the literature, including the Salience, the Infomax and the Variational Free Energy, under a single information theoretic construct, namely the view-based Information Gain. Pursuing this analytic track, two original action selection metrics named the Information Gain Lower Bound (IGLB) and the Information Gain Upper Bound (IGUB) are then proposed. Showing either a conservative or an optimistic bias regarding the Information Gain, they strongly simplify its calculation. An original fovea-based visual scene decoding setup is then proposed, with numerical experiments highlighting different facets of artificial fovea-based vision. A first and principal result is that state-of-the-art recognition rates are obtained with fovea-based saccadic exploration, using less than 10% of the original image's data. Those satisfactory results illustrate the advantage of mixing predictive control with accurate state-of-the-art predictors, namely a deep neural network. A second result is the sub-optimality of some classical action-selection metrics widely used in the literature, that is not manifest with finely-tuned inference models, but becomes patent when coarse or faulty models are used. Last, a computationally-effective predictive model is developed using the IGLB objective, with pre-processed visual scan-path read-out from memory, bypassing computationally-demanding predictive calculations. This last simplified setting is shown effective in our case, showing both a competing accuracy and a good robustness to model flaws.

Dataset Information

A model of collective behavior based purely on vision.

Publications

A model of collective behavior based purely on vision.

Similar Datasets

OmicsDI is part of the ELIXIR infrastructure

Tweets