Summary: New findings indicate that flies use predictive processing in their visual system to execute rapid evasive maneuvers, suggesting prediction may be a widespread strategy across animal nervous systems for supporting fast behavioral responses.
Source: University of Chicago
Researchers at the University of Chicago report that flies anticipate changes in their visual environment to carry out swift escape maneuvers. By relying on predictive information, the fly visual system bridges processing delays and enables rapid behavioral adjustments—an approach that may be common across animal nervous systems.
The study was published on May 20 in PLOS Computational Biology.
Sensory systems collect information about the world, but processing that information takes time. When the environment changes faster than neural processing can keep up, animals face a challenge: how to respond accurately and quickly to threats despite intrinsic neural delays.
“This is crucial in predator–prey interactions,” said senior author Stephanie Palmer, PhD, Associate Professor of Organismal Biology and Anatomy at the University of Chicago. “For a fly, nearly everything is a potential predator, and it must respond in milliseconds. We wanted to understand how flies perform precise evasive actions when sensory feedback arrives too slowly to guide those actions in real time.”
To address this question, the team used an interdisciplinary approach that combined previously recorded behavioral trajectories with detailed computational models of the fly visual circuitry. “This project benefited from open science: we applied theoretical and computational tools to high-resolution behavioral data collected by other investigators,” Palmer said.
First author Siwei Wang, PhD, a postdoctoral researcher in Palmer’s lab, extended earlier theoretical work on motion encoding in the fly visual system to test whether that same circuitry could produce predictions that span the system’s processing delay. “I developed a model to test whether predictive encoding in the early visual network could explain how flies adjust their evasive maneuvers based on the initial detection of a threat,” Wang explained.
Using precise maps of neuronal connectivity in the blowfly visual system, the researchers simulated visual responses while replaying the recorded behavioral dataset. They compared the fly’s actual predictive signals with what an optimal predictor would produce, then dissected the model to find which circuit elements were essential for making those predictions.
The analysis showed that visual information flows through a bottleneck in the fly brain: the sensory stream contains more raw data than the nervous system can fully process, so some information must be discarded. But not all unused data is expendable—some features are necessary for accurate short-term predictions that guide behavior.
Key to this selective filtering are structures called axonal gap junctions—direct electrical connections between neurons. The team found that these gap junctions implement an efficient form of the information bottleneck, preserving prediction-relevant signals while discarding redundant or irrelevant data.
The researchers also identified a specialized subpopulation of vertical motion-sensitive (VS) neurons that both encode predictive information and directly innervate the flight steering motor center. This direct sensory-to-motor pathway suggests how predictive signals can rapidly influence steering commands during an evasive flight maneuver, allowing the fly to fine-tune its trajectory even while additional sensory feedback is still being processed.
These mechanistic insights—the role of axonal gap junctions, the information bottleneck, and the direct link from predictive neurons to motor centers—offer a concrete model for how sensory prediction can be transformed into rapid motor responses. Such principles may apply broadly across species that must act faster than their sensory processing delays permit.
“Opening the fly’s neural ‘black box’ revealed design principles that likely generalize to other nervous systems,” Palmer said. The team plans to look for comparable prediction-driven circuits in other animals to test how widely these strategies are conserved.
Beyond basic neuroscience, the findings may have broader implications. Understanding how simple neural circuits implement prediction could inform models of human brain function and even provide insights into disorders where predictive processing breaks down, such as Alzheimer’s disease. Wang cautioned that translating fly results to humans will require much more work, but noted that these studies lay important groundwork for future research.
Funding: The study, “Maximally efficient prediction in the early fly visual system may support evasive behaviors,” received support from the Gatsby Charitable Foundation; the Max Planck–Hebrew University Center for Sensory Processing of the Brain in Action; the National Science Foundation CAREER Award 1652617 and Center for the Physics of Biological Function (PHY-1734030); and the National Institutes of Health (R01EB026943). Additional authors include Idan Segev of the Hebrew University of Jerusalem and Alexander Borst of the Max Planck Institute of Neurobiology.
About this neuroscience research news
Source: University of Chicago
Contact: Alison Caldwell – University of Chicago
Image: The image is in the public domain
Original Research: Open access.
“Maximally efficient prediction in the early fly visual system may support evasive flight maneuvers” by Siwei Wang, Idan Segev, Alexander Borst, Stephanie Palmer. PLOS Computational Biology
Abstract
Maximally efficient prediction in the early fly visual system may support evasive flight maneuvers
Sensory prediction compensates for inherent processing delays, yet the mechanisms linking prediction to natural behavior remain underexplored.
We show that despite a 20–30 ms intrinsic delay, the blowfly’s vertical motion-sensitive (VS) network achieves near-optimal prediction. This predictive encoding enables the fly to adjust short, complex evasive maneuvers according to its initial self-rotation detected at the moment of threat.
By combining a comprehensive behavioral dataset with detailed compartmental models of the VS network, we demonstrate that axonal gap junctions are essential for achieving efficient prediction.
During evasive flight, a VS neuron subpopulation that directly connects to the neck motor center carries predictive information about future self-rotation, which could be critical for ongoing flight control.
These findings reveal a sensory-motor pathway that links early sensory prediction to rapid behavioral output.