| Summary: | <p>In the brain of the fruit fly <em>Drosophila melanogaster</em>, dopaminergic neurons are thought to convey teaching signals to synapses of the mushroom body during associative olfactory learning. Although rewarding dopaminergic neurons in mammals appear to encode prediction error, whether the equivalent neurons in Drosophila have a similar function remains unknown. Anatomically distinct subsets of dopaminergic neurons reinforce water and separate properties of sugars: sweet taste reinforces short-term memories while nutrient input is usually required for long-term memories. This thesis investigates dopaminergic reinforcement at the physiological level using two-photon in vivo calcium imaging of tethered flies during feeding, odour presentation, and appetitive conditioning.</p> <p>One subset of dopaminergic neurons reinforces short-term sugar memory in starved flies and short-term water memory in dehydrated flies. During sugar-water droplet feeding, dopaminergic neurons show a calcium response that reflects salience and whose magnitude increases with sweetness of the droplet. In starved flies, dopaminergic neurons show differential responses to water and sugar, whereas in dehydrated flies, the neurons show uniform feeding responses. Moreover, sucrose-evoked responses in satiated flies are substantially lower than in either deprivation state. Such findings suggest a deprivation state-dependent valuation of reward not explained by current models of reward prediction error.</p> <p>Feeding responses in the same dopaminergic neurons or other rewarding subsets involved in sugar reinforcement do not increase with the nutritious content of the droplet. However, feeding rapidly inhibits aversively reinforcing dopaminergic neurons that also control the state-dependent expression of sugar memories. Spontaneous calcium oscillations in the aversive reinforcing neurons become more frequent after feeding on sugars specifically, and their amplitude after feeding appears to be diminished by a possible rapid nutrient signal. Intriguingly, rewarding dopaminergic neurons also show substantial odour-evoked calcium responses that reflect both general and specific odour novelty; such activity is not yet accounted for in existing models of associative learning in Drosophila. </p> <p>Rewarding dopaminergic neurons recorded during appetitive conditioning do not show acquired predictive capacity after training as expected from current models of reward prediction error. Although it remains unclear whether Drosophila dopaminergic neurons are inhibited by the unexpected omission of reward, one subset in particular appears to be excited by the unexpected presentation of reward. Taken together, these findings show that rewarding dopaminergic neurons in Drosophila may encode value, salience, novelty, and positive prediction error during associative learning.</p>
|
|---|