Cookies on this website

We use cookies to ensure that we give you the best experience on our website. If you click 'Accept all cookies' we'll assume that you are happy to receive all cookies and you won't see this message again. If you click 'Reject all non-essential cookies' only necessary cookies providing core functionality such as security, network management, and accessibility will be enabled. Click 'Find out more' for information on how to change your cookie settings.

Making decisions in the real world is challenging because choices made now influence what choices will be available in the future. As the number of steps in a sequence of choices increases, the potential number of paths through a decision tree increases exponentially. How are we able to make adaptive decisions in the face of such overwhelming complexity? One idea is that the brain uses shortcuts, or heuristics, to reduce computational demands. Previously we provided evidence for the existence of one such heuristic, "pruning", that entails avoiding even considering entire branches of the decision tree that begin with a large negative outcome, regardless of subsequent outcomes. We found that planning was profoundly impaired when the optimal decision entailed initially accepting a large negative outcome (Huys et al 2012 PLoS Computational Biology 8(3):e1002410), and computational modelling showed that this bias could not be explained by other influences such as loss aversion.
 
In this talk I will outline our recent work examining the neural basis of pruning. In a neuroimaging study we first demonstrated the ubiquity of pruning behaviour - planning was impaired in every participant when the optimal sequence entailed accepting a large loss, which computational modelling confirmed was driven by pruning. Categorical neuroimaging analyses revealed that avoiding planning through large losses recruited structures implicated in Pavlovian conditioning and depression, specifically the subgenual anterior cingulate cortex (sgACC) and amygdala. Using our computational model to infer the influence of pruning on a trial-by-trial basis, estimated as the Kullback-Leibler divergence between action probability distributions arising from models with and without pruning, we confirmed the central role of sgACC. Receiving large losses activated the anterior insula and deactivated the ventral striatum; interestingly, responses in the former, but not the latter, were strongly associated with the degree of pruning behaviourally across participants. Finally, in a separate study we investigated pruning in individuals enrolled in a trial of a single-dose of 3,4-methylenedioxymethamphetamine ("ecstasy"). This drug releases serotonin for several hours, but is thought to deplete reserves during the succeeding days, during which depressive symptoms can be temporarily elevated. During this period we found that large negative outcomes had less of a detrimental influence on planning; however, computational modelling could not confirm that this was related to a decrease in pruning specifically. The results will be discussed with reference to a contemporary theoretical framework that relates Pavlovian behavioural inhibition to serotonin and depressive symptoms.