In 2001, Miller and his co-author Jonathan Cohen published a highly influential theory (the 5th-most cited paper in neuroscience) of how the prefrontal cortex (PFC) exerts “executive” control over brain function. The PFC acts like an air traffic controller. It acquires and uses top-down information (like rules, concepts, and knowledge) to direct the flow of activity in the rest of cortex. A key property was that the neurons are multifunctional.
For most of the history of neuroscience, the field’s doctrine held that individual neurons had individual roles, but measurements made in the lab of Earl K. Miller have revealed that neurons in the brain’s prefrontal cortex are multifunctional. Beginning in 2000, several studies in Miller’s lab and broader collaborations showed that the same neurons can signal different information in different tasks, indicating that they are capable of taking part of different ensembles working on different problems (i.e. they exhibit “mixed selectivity”).
In an especially important paper in 2013 with colleague Stefano Fusi, Miller also found evidence that the mixed selectivity of neurons provides the brain with the computing power needed for high-level cognition. Neurons that can be involved in multiple tasks and that can incorporate multiple dimensions of a task (e.g. respond to a specific image only if it follows another) can compose circuits that are able to flexibly and quickly adapt to execute new tasks, for instance. Notably, in the study they showed that when the multidimensional nature of neurons fails, animals make more errors on cognitive tasks in which they’ve been trained.