Data-driven modelling of social forces and collective behaviour in zebrafish

Abstract

Zebrafish are rapidly emerging as a powerful model organism in hypothesis-driven studies targeting a number of functional and dysfunctional processes. Mathematical models of zebrafish behaviour can inform the design of experiments, through the unprecedented ability to perform pilot trials on a computer. At the same time, in-silico experiments could help refining the analysis of real data, by enabling the systematic investigation of key neurobehavioural factors. Here, we establish a data-driven model of zebrafish social interaction. Specifically, we derive a set of interaction rules to capture the primary response mechanisms which have been observed experimentally. Contrary to previous studies, we include dynamic speed regulation in addition to turning responses, which together provide attractive, repulsive and alignment interactions between individuals. The resulting multi-agent model provides a novel, bottom-up framework to describe both the spontaneous motion and individual-level interaction dynamics of zebrafish, inferred directly from experimental observations.

The sensory basis of schooling by intermittent swimming in the rummy-nose tetra (Hemigrammus rhodostomus)

“…Schooling is a form of shoaling where the fish move with a similar heading [16], measured as the group polarization. The polarization of a group may be achieved indirectly as a consequence of fish moving towards other fish [4], or results from actively seeking alignment with neighbours [17,18]. Pollock swim more closely together when the lateral line system is compromised, which supports the hypothesis that flow sensing mediates repulsion and vision is the modality used for attraction [8].…”

The zebrafish subcortical social brain as a model for studying social behavior disorders

“…Researchers have modeled individual behavioral rules in response to the motion of a social stimulus fish using theoretical frameworks based on the Bayesian decision theory (Box 2) (Arganda et al, 2012) and transfer entropy (Box 2) (Porfiri and Ruiz Marín, 2017). Other studies first improved continuous tracking of individuals and then computationally modeled pairwise interactions using the optimal control theory (Laan et al, 2017), deep attention networks (Heras et al, 2018), transfer entropy (Butail et al, 2016) and other data-driven methods (Zienkiewicz et al, 2018) to reveal how pairs of individuals attract, repulse and align with each other.…”

A Comparison of Individual Learning and Social Learning in Zebrafish Through an Ethorobotics Approach

“…The replica was controlled by programming sequences of 2D coordinates and sending them from the PC to another Arduino microcontroller. The sequence was generated by implementing a stochastic mathematical model of zebrafish swimming, which we have established in our previous work (Mwaffo et al, 2015Zienkiewicz et al, 2018). The model captures the typical burst-and-coast swimming style of zebrafish, where sudden tail beats are followed by longer coasting phases.…”

“…The demonstration by the replica entailed the following steps. At the beginning of each trial, the replica interacted with the focal subject for 30 s, following a trajectory generated via our stochastic model of zebrafish locomotion (Mwaffo et al, 2015Zienkiewicz et al, 2018) with an attraction point at the center of the cylinder that housed the experimental subject. This resulted in the replica swimming in the focal region, while frequently approaching the subject and "wall kissing" the cylinder.…”