ABSTRACT
The central problem in insect olfaction is to identify the quality of an odor stimulus impinging on the antennae of an individual, given that odor stimuli are often borne on highly turbulent airflows and sometimes are embedded in noisy odor backgrounds. From the literature, it is evident that the response of olfactory receptor cells of olfactory generalist insects (e.g., worker honey bees) are highly nonlinear in response to changes in odor concentration and odorant blend (synergistic and, especially, inhibitory phenomena are ubiquitous). Thus it appears beyond the capabilities of linear filters to extract general odor quality information from receptor input. The question is: what sort of neural network is required, given the spatio-temporal structure of natural odor stimuli and empirical measured response behavior of receptor neurons in the insect antenna, to successfully extract odor quality information? Here I will formalize some of the essential features of the odor perception problem, as well as review what is know about the network architecture of olfactory neurons in the antennal lobe of insects. I will contrast critical differences between pheromonal and non-pheromonal detection subsystems and identify physiological and anatomical issues that need to be resolved to gain a fuller understanding of receptor input integration and feature extraction in insects.
