How Brain Memory Uses Complex Sight for Navigation

Greg Howard
24th June, 2025

How Brain Memory Uses Complex Sight for Navigation

This study models the experimental finding that navigating wood ants (Formica rufa) use the fractional position of mass of a visual shape (a) by using a bilateral neural network (b) to test if this higher-order processing emerges from balancing separate left and right visual memories.

Image adapted from: Filippi et al. / CC BY (Source)

Key Findings

  • A study from the University of Sussex and UC Santa Barbara found that ants' complex visual navigation, like using how much of a shape is on one side, emerges from their brain's two-sided structure
  • This suggests that ants' brains don't need complex calculations for "higher-order" visual cues; instead, these abilities naturally arise from how visual information is processed across brain halves
Insects, particularly ants, are renowned for their remarkable navigational abilities, relying heavily on memories of their visual surroundings to find their way through complex environments. Foragers embark on long, visually guided journeys to gather food and then navigate back to their nest. While we understand the fundamental neural pathways involved – visual information travels from the Optic Lobes to the Mushroom Bodies, which are key centers for memory storage in the insect brain – the precise nature of how visual scenes are represented and used for navigation has remained largely unknown. Previous research has established the incredible precision with which ants use visual landmarks. For instance, studies on Australian desert ants, Melophorus bagoti, showed they develop highly accurate, habitual routes between their nest and food sources[2]. These route memories are acquired quickly and are so robust that ants can rejoin their established paths at any point and follow them accurately, regardless of their internal 'path-integration vector' – an internal sense of direction and distance traveled[2]. This highlights the critical role of visual memory, independent of other navigational cues. It has often been suggested that ants utilize "higher-order" visual information for navigation. This refers to features extracted not just from a single point, but from the overall arrangement and relationships within an entire visual scene. One such feature is the "fractional position of mass" (FPM), which describes the proportion of a visual shape that appears to the left or right of an ant's central line of sight. The question then arises: how might an ant's brain compute such a complex feature? Recent research from the University of Sussex and University of California Santa Barbara[1] has shed new light on this puzzle. This study explored whether the apparent use of FPM by ants could simply be an emergent property of the insect brain's bilateral organization, even if the visual input itself is processed in a relatively simple "retinotopic" way. Retinotopic processing means that visual input from different parts of the eye maps directly to corresponding areas in the brain, much like a direct projection of the visual field. The bilateral organization refers to the brain having two distinct halves, or hemispheres, each processing information. To investigate this, the researchers developed a simplified model of ant visual memory, constrained by the known neuroanatomy and information processing characteristics of the Mushroom Bodies. This model aimed to simulate how visual information might be stored and retrieved. Their key finding was that such a bilaterally organized memory model can, remarkably, implicitly encode the FPM that ants appear to learn during training. This means the brain might not need a dedicated, complex module to calculate FPM; instead, this "higher-order" information could naturally arise from the way visual inputs are processed and stored across the two hemispheres. The model demonstrated that by balancing the "quality" of the memory match across the left and right hemispheres, it could successfully retrieve the direction dictated by FPM. Crucially, this held true even when the model was presented with entirely new, unfamiliar shapes, mirroring observations in real ants. This suggests a powerful and flexible mechanism. Furthermore, the results were largely independent of specific model parameters, reinforcing the idea that this processing might be a fundamental consequence of the neural circuit's structure. This new understanding builds upon and ties together earlier concepts of ant navigation. For instance, previous models proposed that ant route navigation could be understood as a "search for familiar views"[3]. In these models, a familiar visual scene would directly specify a familiar direction of movement, and this search could be performed using simple scanning behaviors observed in ants. The current study provides a deeper, mechanistic explanation for how these "familiar views" might be represented and processed internally, suggesting that the "familiarity" isn't just about matching a stored image, but involves a more abstract, emergent property like FPM. Moreover, the findings resonate strongly with research that challenged the long-held assumption that ants navigate by storing and retrieving discrete "snapshots" or sequences of individual views[4]. Instead, that work suggested that ants might gather information from all experienced views into a single, integrated memory network, using it to determine the most familiar heading at any given location. The current study's demonstration that FPM can emerge from bilateral processing, rather than being explicitly computed, supports this idea of a more integrated, continuous visual memory system. It suggests that the complex visual guidance observed in ants, which allows them to follow long, idiosyncratic routes and accurately pinpoint locations in varied environments, might stem from surprisingly parsimonious mechanisms. Rather than requiring dedicated, complex processing modules, some aspects of "higher-order" visual processing may simply emerge from the fundamental architecture of the ant brain's neural circuits.

Animal Science

References

Main Study

1) Lateralised memory networks may explain the use of higher-order visual features in navigating insects

Published 23rd June, 2025

https://doi.org/10.1371/journal.pcbi.1012670

Related Studies

2) Idiosyncratic route-based memories in desert ants, Melophorus bagoti: how do they interact with path-integration vectors?

Journal: Neurobiology of learning and memory, Issue: Vol 83, Issue 1, Jan 2005

3) A model of ant route navigation driven by scene familiarity.

https://doi.org/10.1371/journal.pcbi.1002336

4) Snapshots in ants? New interpretations of paradigmatic experiments.

https://doi.org/10.1242/jeb.082941


Related Articles

An unhandled error has occurred. Reload 🗙