Check out this National Public Radio piece on the debate about our new model of working memory:
Neuroscientists Debate A Simple Question: How Does The Brain Store A Phone Number?    I have a few follow-up points.

1. In the piece, Christos Constantinidis says: “The problem with the theory is that so far there has been no experimental evidence linking this critical variable with behavior,”

This is false.  We have a paper entirely devoted to experimental evidence linking the theory to behavior:
Lundqvist, M., Herman, P. Warden, M.R., Brincat, S.L., and Miller, E.K. (2018) Gamma and beta bursts during working memory read-out suggest roles in its volitional control. Nature Communications. 9, 394   View PDF

2. Christos also says:  Miller’s contention that working memory is linked to long-term memory seems at odds with doctors’ experience with patients whose brains have been injured.

This totally misses the point.  Working memory can exploit some of the same mechanisms as long-term memory (synaptic weight changes) while at the same time rely on different brain areas.  The fact that you can have brain damage that disrupts working memory without disrupting long-term memory is completely irrelevant.

3. Most importantly, the people who support the old model of persistent activity have not done the crucial test. All the evidence for the old model of persistent activity averages activity across trials.  You cannot do this.  Averaging creates an illusion of persistence.  You must examine activity on single trials.   Unless you do that, you are not addressing the issue.

If you want to read about our new working memory, check out this paper:
Miller, E.K., Lundqvist, L., and Bastos, A.M. (2018) Working Memory 2.0  Neuron, DOI:  View PDF

About the Author

The Miller Lab uses experimental and theoretical approaches to study the neural basis of the high-level cognitive functions that underlie complex goal-directed behavior.