We have the unique opportunity to give a definitive answer as to how associative learning is integrated with action in a biological brain by combining neurogenetics for functional dissection with modelling in a one-to-one description of neural computation in the larvae linked to behavioural control. Parts of this pathway are already well described, in particular the olfactory receptor – olfactory bulb – mushroom body circuit that appears crucial to olfactory learning  . Further information is rapidly emerging regarding the gustatory pathways , reinforcement neurons , and motor circuits , although the relevant premotor and motor circuits for learnt behaviour are not yet clear.
Still needed is a closed loop control description that can be interfaced to an agent to capture/predict innate behaviour and its alteration during learning. A preliminary spiking neural model that captures what is known or hypothesised so far about the closed loop control will be refined through neurogenetic experimentation throughout the project (WP4 Objective 4.1 – To develop a complete one-to-one model of the neural circuit of learnt behaviour expression in larva).
We will be able to test our understanding of the role of each part of this circuit during olfactory-gustation association experiments by blocking, or more excitingly, substituting ‘virtual’ stimuli using optical and thermal activation of target neurons (WP1 Objective 1.3 – To use genetic methods to dissect the neural circuits underlying acquisition and expression of learning in this paradigm).
Our initial focus will be on identified olfactory receptors, gustatory receptors, and dopaminergic and octopaminergic interneurons in the reinforcement pathways. As well as allowing confirmation of their functional roles, this approach enables presentation of arbitrary precise temporal relations of stimulus and reward, linked or dissociated from each other and from action, which can be used to tease apart the precise mechanisms of associative acquisition. The same methods will then be used to target the less well understood but crucial function of the mushroom body extrinsic (output) neurons. In what
sense do these represent the outcome of learning: as direct controllers of attractive or aversive behaviour; as a ‘valuation’ of the stimulus that interacts with the context and motivation to determine control; or as an explicit prediction of the reinforcer? (WP1 Objective 1.4 – To understand how the larval brain integrates innate and learnt behavioural tendencies)
To complement these methods, and to inform and validate the model, we will also develop methods to read out the activity of the same target neurons in a behaving animal using biofluorescence. We can take advantage of our ability to genetically target known neurons to obviate the requirement of precise spatial localisation for identification, which opens up the opportunity to use these methods in the naturally behaving animal, experiencing and acting under the same conditions as in the behavioural experiments, so that we can synchronise brain activity and behaviour (WP3 Objective 3.1 – Develop bioflourescence methods to read activity of identified neurons during behaviour).
These approaches will enable a tight interaction of modelling and experimentation that will accelerate progress towards a complete description of a behaving brain at the single neuron level.
2 – First neural model implemented and interacting with agent simulation (month 7)
4 – Bioluminescence in static animal (month 13)
7 – Results from optogenetically controlled conditioning (month 19)
10 – Biofluorescence in moving animal (month 25)
11 – Neural model revised from data and predictions (month 26)
15 – Tested predictions of model by control and monitoring of extrinsic neurons (month 36)