Cookies on this website
We use cookies to ensure that we give you the best experience on our website. If you click 'Continue' we'll assume that you are happy to receive all cookies and you won't see this message again. Click 'Find out more' for information on how to change your cookie settings.

What is the brain’s mechanism for decision-making? In spite of all we know, this critical cognitive function has yet to be captured by a unifying formalism, akin to E = mc2. This is just becoming possible.

In this talk we will introduce a class of single-equation probabilistic procedures that make choices based on sets of simultaneous noisy spike trains, thereby joining neural signalling in its native format with decisions at the behavioural level. We go on to show that, when making choices based on the statistical structure present in the sensory cortex of the monkey, such procedures –ideal Bayesian observers for two alternative settings- render reaction times far shorter than those of monkeys, under the same conditions. This predicts that animals lose some of the information in this structure. By corrupting the structure to exactly match their information loss, we recover algorithm reaction times that tightly match those of monkeys. Finally, we explain how our algorithm maps onto the recurrent architecture of the cortico-basal-ganglia-thalamo-cortical loops, a key substrate for decision formation. This enables us to show that the time course of the computations that map onto sensory-motor cortex, qualitatively matches that of the neural activity recorded from the lateral intra-parietal area of the macaque during decision formation. In addition, it predicts the course of recordings at the output nuclei of the basal ganglia and thalamus, which have yet to be performed. We argue that this single-formula, exact-inference framework, capable of simultaneously matching the behaviour, anatomy and existing neural data is a firm step towards THE mechanism.

Host: Chris Summerfield