State of the art: Drosophila melanogaster is a small fly that has been used as one of the major biological research tools for around 100 years; much of what we know about genetics, neuroscience and development, and now learning, comes from this animal. The lifecycle of Drosophila
melanogaster comprises four stages: egg, larva, pupa, and adult. At the larval stage, flies centuplicate their weight in only 6 days.
Thus, larvae are motivated by a single objective: localizing substrates rich in proteins and sugars to permit their development through to the adult stage. This includes a range of
innate orienting responses to gustatory and olfactory cues, as well to temperature, gravity, light and tactile signals. For odour cues the orientation response is not ultimately about odour but about finding or avoiding odour sources; as a consequence, olfactory processing includes pathways with an enormous potential to discriminate odours and to flexibly attach acquired meaning to them, leading to altered behaviour. LIN recently established a robust paradigm for odour learning in larval Drosophila  reviewed in . The paradigm presents an odour with a sugar reward and measures the consequential change in preference for the odour (Figure 1); alternatively, low-concentration salt can be used as gustatory reward . Larvae have also been shown to learn with aversive tastants such as high-concentration salt or quinine  with alternative reinforcers such as shock  or vibration , and to learn about the association of visual cues with gustatory reinforcement . The odour-sugar paradigm is the main focus of the current proposal because it has already been analysed in some detail with respect to its neurogenetic underpinnings e.g. . Importantly in the current context, we have shown that the association between the odour and internal reward signal happens in the socalled mushroom bodies (MB, see section 1.2.4), in line with a rich experimental and modeling literature on learning in adult flies as well as other insects  . Also, the paradigm has been used to test generalisation between different odours as well as between different concentrations of the same odour, and for characterizing the learning of odour mixtures .
Baseline: There is clear evidence that the change in the larval response to odours after pairing with appetitive or aversive tastants is not simply assimilation of the unconditioned response to the conditioned stimulus, and that the increased attraction towards the odour represents an action in search for food, rather a mere response towards an appetitively ‘charged’ odour. LIN and UEDIN have shown a dramatic dissociation between learnt behaviour towards the odour versus innate behaviour towards the reinforcer . Larvae find some high concentrations of salt aversive; however, pairing such a repulsive salt concentration with an odour can lead to increased attraction to the odour.
A similar effect was demonstrated with sugar, which becomes aversive at very high concentrations, but still acts as reward . For quinine, on the other hand, low concentrations which produce no innate avoidance response can still be effective reinforcers leading to aversion to the odour .
Further results from LIN (Fig 2) show larvae with identical training (A) show different behaviour (B) depending on the test conditions: they show increased attraction to the odour only if the level of sugar in the test condition is less than that paired with the odour in training. This effect is not due to direct effects of sugar presence (C) on behaviour to odour, or vice versa (D). On the other hand if test conditions are the same but training differs (B, far right group), larvae again only show the response to odour if the test level is less than that paired with the odour in training. It appears outcome expectation, and not the value of the memory trace, is the immediate cause of conditioned behaviour , as reviewed in .