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

Area of Science:

• Neuroscience
• Animal Behavior
• Computational Biology

Background:

• Insects, particularly ants, rely on visual memories for spatial navigation and foraging.
• The neural pathways for visual memory storage (Optic Lobes to Mushroom Bodies) are known, but the representation of visual scenes for navigation remains unclear.
• Previous studies suggest ants use "higher-order" visual information, such as the fractional position of mass (FPM), for route learning.

Purpose of the Study:

• To investigate if the bilateral organization of the insect brain, specifically the Mushroom Bodies, can implicitly encode the fractional position of mass (FPM).
• To explore how a simple, retinotopic visual representation within a bilaterally organized model might explain ants' use of FPM for navigation.
• To determine if emergent properties of neural circuits, rather than discrete modules, underlie higher-order visual processing in ants.

Main Methods:

• A computational model was developed, constrained by the known neuroanatomy and information processing of insect Mushroom Bodies.
• The model assumed a simple "retinotopic" visual representation.
• The model was trained to learn and retrieve spatial information based on visual cues, specifically focusing on FPM.

Main Results:

• The bilaterally organized memory model successfully and implicitly encoded the FPM learned during training.
• Balancing memory match quality between the left and right hemispheres enabled the model to retrieve FPM-defined directions.
• The model accurately predicted FPM-based navigation with novel shapes, mirroring ant behavior, and results were robust to parameter variations.

Conclusions:

• The bilateral organization of neural circuits can implicitly encode higher-order visual features like FPM, crucial for spatial navigation in ants.
• Emergent properties from neural circuit structure, rather than specialized modules, may explain aspects of higher-order visual scene processing in insects.
• This model provides a neuroanatomically constrained explanation for how ants might compute and utilize FPM for navigation.