Conference on Cognitive Computational Neuroscience

Despite having a brain with less than a million neurons, honey bees demonstrate higher-order cognitive functions, being able to solve non-elemental forms of learning and abstract tasks. Experimental studies showed that honey bees successfully perform trace and delay conditioning tasks, associating a conditioning stimulus (CS) with an unconditioned stimulus (US). In humans, trace conditioning requires higher-order functions such as attention and memory, and it has been even suggested to relate to conscious perception. Here, taking inspiration from the honey bee's neurobiology, we investigated the cognitive architectural ingredients necessary to perform trace conditioning tasks. Specifically, we modeled a population of 560 Kenyon cells in the honey bees' mushroom bodies (MB), a structure involved in associative learning and memory. We demonstrated that MB neurons learn to associate an olfactory CS with a sucrose US via spike-time-dependent plasticity (STDP). Additionally, we modeled an attentional mechanism that allows disregarding distractors stimuli, in line with recent experimental findings. Our results matched the experimental observations on olfactory trace conditioning in honey bee, corroborating our approach. Overall, our results give new insights into the neural mechanisms involved in trace conditioning, providing a computational benchmark to test future predictions and unveil the mechanisms of these miniature cognitive architectures.