Understanding the processes engaged in the brain to allow us making such a choice is a central question in neuroscience of decision-making and many aspects of it remain unclear. A decomposition of this process might help us to understand the involved mechanisms. First we need to assign what we will call a ‘subjective value’ to each option – i.e. the quantification of how much we like each of these options. Then, we need to compare those values to finally being able to select one of them.

In Lopez-Persem et al, Plos Computational Biology 2017, we investigated and compared three behavioral ways to have an access to these ‘subjective values’:

• a rating task - subjective report on how much a reward is desired

• an effort task - physical effort investment to obtain a reward

• a binary choice task - choices between two rewards

We found that subjective values are relatively robust to the way they are elicited.

In Lopez-Persem et al. Nature Neuroscience 2020, we investigated the properties of the Brain Valuation System using deep brain electrophysiological recordings in humans. We have demonstrated that subjective value could be decoded:

• in pre-stimulus activity (the higher the baseline is, the more we will like an item)

• for various categories of items (genericity)

• even during a distractive task (automaticity)

• as both linear and quadratic signals (encoding both value and confidence).

Thus, our findings specify key functional properties of neural value signals which might provide insights into human irrational choice behaviors.

For more details, see the article or the blog post summarizing it.

In Lopez-Persem et al. eLife 2016, we investigated how this brain network was involved during binary choices. We specifically investigated whether prior preferences defined at the category level (savory versus sweet food for example) would define a default policy towards one or the other item (oyster versus macaroon). At the behavioral level, we found that prior preferences induce a bias that leads participants to choose the default option more often (and faster) than its value would predict. At the neural level, we found that prior preference influenced the BVS baseline activity and that decision value was expressed by the vmPFC in a default versus alternative framing.