Flies neglect food and endure shocks to seek a dopamine reward : Waddell Group paper published in Nature

Using optogenetic activation, the team discovered that they could generate olfactory associations that starved flies would seek while neglecting food or enduring electric shocks. This was not observed when flies were trained to associate an odour with the activation of other neurons in the brain or with a natural reward such as sugar.  

The researchers discovered that flies took risks to endure shock while seeking reward because the dopamine neurons that ordinarily signal electric shock punishment were functionally impaired by prior activation of the reward-encoding dopamine neurons. This revealed antagonism between reward-encoding and punishment-encoding dopamine neurons in the brain. 

Ordinarily, reward-encoding dopamine neurons are thought to send teaching signals to the mushroom body to reinforce olfactory associative learning. However, starved flies sought sugar less even after the reward-encoding dopamine neurons were activated in the absence of odour. This suggested that the dopamine neurons also convey satiety-like ‘demotivational signals’. 

Since flies ordinarily do not neglect food or endure shock to seek reward, the team reasoned that the activity of these dopamine neurons must ordinarily be tightly controlled. In the paper, the authors present physiological and anatomical evidence that this is indeed the case.  

Calcium imaging of these dopamine neurons using two-photon microscopy revealed that the dopamine neurons convey signals that are specific to both reward type and the physiological state of the fly. The researchers also observed calcium responses that resembled teaching signals and satiety-like signals in these neurons.  

Moreover, using the latest connectome data of the fly hemibrain, the authors found that the reward-encoding dopamine neurons (approximately 60 in total) receive inputs from over 1700 neurons from all over the brain (visualized in the connectome image below). This is over 25 times as many inputs, suggesting that these dopamine neurons have elaborate input controls. 

With all this evidence, Professor Scott Waddell and Dr Kristijan Jovanoski propose that these heterogeneously rich reward-encoding dopamine neurons are ordinarily tightly controlled by upstream neurons that convey reward type and physiological state. Optogenetically activating these dopamine neurons bypasses their elaborate input control and destroys the reward-specificity and state-specificity of their signalling. Consequently, flies seek a non-specific reward that is greater than the sum of individual rewards, leading to supranormal reward seeking despite electric shocks or physiological needs. 

Given that there are many parallels between the dopamine neurons of flies and mammals, the authors propose that the mechanisms discovered here may similarly apply towards understanding compulsive reward-seeking behaviour and addiction in mammals.