Excited to announce that my PhD research is now out in @Nature. We characterise a medial-frontal network for novel choice in primates involving a grid representation of value https://rdcu.be/cc9ZZ  W/ Matthew Rushworth @MKFlugge @davidefolloni @LennartVerhagen @jerome_sallet 1/8

Have you always wondered why electrophysiological studies of value-guided decisions in primates find consistent effects in orbitofrontal cortex (OFC), while human fMRI studies typically identify value effects in medial frontal cortex (MFC)? We may have found an explanation... 2/8

Lab animals routinely make thousands of choices among a small set of options. But in real life people often face new items they have never tried before, requiring value to be computed on the fly. Here we designed a task to get monkeys to make new choices, just like humans. 3/8

The macaques chose between pairs of stimuli, each characterised by two attributes: magnitude and probability of reward. Crucially, some attribute combinations were pre-trained for months (in grey in the right panel), others were novel at the time of fMRI testing (in white). 4/8

We searched for a value comparison effect in the BOLD signal (chosen minus unchosen option value). As expected, such a signal was present in OFC and equally for all trials, but it was strikingly different in MFC when choosing among unfamiliar compared to familiar options. 5/8

How does MFC represent novel 2-dimensional value (magnitude and probability)? We presented options one at a time, so that the monkeys could “navigate” this space, and we observed a hexagonal symmetry in the MFC fMRI activity, analogous to the grid encoding of physical space. 6/8

Finally, we transiently disrupted MFC with a new focused ultrasound method. This caused choice behaviour to rely more on the two attributes considered separately, and less on their interaction, proving MFC’s key role in the value integration process underlying novel choice. 7/8

Together, these results suggest the existence of a medial frontal circuit in the primate brain causally involved in making novel choices – choices based on attribute integration rather than previous direct reinforcement – and grid-like encoding of such multi-attribute spaces. 8/8