Though I “supervised” Mark Stokes' PhD, the direction of influence was never entirely clear. On a memorable evening in Havana, in the second year of his studies, Mark sat me down at a bar and suggested we should “give these new multivariate fMRI methods a try.” Though these methods are now very standard, at that time there was a lot to consider, and Mark more or less single-handedly led the development of multivoxel pattern analysis in the Cambridge community. A few years later, now well established in Oxford, Mark approached me again and suggested that similar multivariate methods might usefully be applied to our neurophysiological data from the behaving monkey. Having learned by now that the best bet was just to sit back and let him go, I was happy to hand over data from a large sample of frontal lobe neurons recorded during a working memory task. Every few...

You do not currently have access to this content.