Back

Context dependent multisensory integration: Mechanosensation depends on luminance for robust performance

Sharma, V. P.; Sponberg, S.

2023-10-02 neuroscience
10.1101/2023.09.29.560224 bioRxiv
Show abstract

Multisensory, goal-directed behaviors are ubiquitous and essential in animals across taxa. Individual sensory modalities may be significantly tuned by environmental context, and yet they must continue to combine and produce multisensory behaviors. This necessitates either behavioral robustness or goal-relevant adaptation in sensorimotor systems. In the case of hover-feeding hawkmoths of the species Manduca sexta, proboscis mechanosensation and vision are known to combine linearly in tracking of moving flowers. This tracking behavior is affected by light-levels, and visual compensation is necessary for robust flower-tracking. Whether the mechanosensory response also adjusts to light-level, in order to produce linear combination across luminance contexts, is not known. We applied control theoretic analyses to study integration of mechanosensory and visual information in flower tracking hawkmoths in the context of high luminance. This was performed using a two-part robotic flower system. Under the conditions of high luminance, we verified the linear combination of mechanosensory and visual systems, and the complementary modulation of these systems. We show that both systems tune their frequency response with changing environmental light level,while keeping their combined error low at relevant frequencies. Thus, the animal adapts to changing conditions, across modalities, even when only one of the modalities (vision) is directly affected. This cross-talk between sensorimotor systems is constrained, and possibly driven, by performance requirements.

Matching journals

The top 4 journals account for 50% of the predicted probability mass.

50% of probability mass above

"Similar papers" are the closest papers from that journal in the model's embedding space. They show what the match is built on, but the ranking comes mostly from a classifier over the whole training set, not from these examples alone.