Working Out Working Memory: A Life’s Pursuit

Q&A with Earl Miller

Working memory is key to our everyday survival — how we communicate, remember what we need to do, learn new things, and generally operate. It is also an aspect of cognition that is disrupted or dysfunctional in almost every neuropsychiatric disorder. Therefore, understanding how working memory works is of vital importance.

For Earl Miller, a cognitive neuroscientist at MIT, breaking down the neurophysiology of working memory has been part of his life’s pursuit. One of his early inspirations in graduate school at Princeton was George A. Miller, an innovative psychologist who was pivotal in advancing the field of cognitive neuroscience. Fittingly, Miller is the latest recipient of the CNS George A. Miller Prize, in recognition of his innovative work to understand working memory.

CNS spoke with Miller, who will be delivering an award talk at the upcoming CNS annual meeting, March 23-26 in San Francisco, CA, and he shared his insights on working memory and the way progress happens in science.

CNS: How did you personally become interested in neuroscience and then cognitive neuroscience specifically?

Miller: My interest in neuroscience started by accident.  I was a pre-med major. I volunteered to work in a neuroscience lab and was immediately hooked. I switched from pre-med to psychology and neuroscience. My mom sat shiva, but she got over it.

As for cognitive neuroscience, I have always been interested in how the mind works. Isn’t everyone?

CNS: In your recent review paper in Neuron, you call working memory a “sketchpad of conscious thought.” What do you mean by that?

Miller: Working memory is what we use when we think.  It is the contents of consciousness, not just the thoughts but what we do with them. Examples of working memory include anything to do with conscious thought — planning, conversation, etc.

CNS:  What major insights on working memory will you be sharing in your CNS 2019 award lecture?

Miller: The classic model of working memory is sustained spiking activity of neurons. You activate an ensemble and it simply stays active. This is correct to a certain level of approximation, one that involves averaging data across trials to get a general picture of neural activity and dynamics. But we and others are finding that when you look at it more closely, in real time, on single trials, there is much more going on. The spiking occurs in brief bursts.  This leaves room for a rhythmic interplay between oscillations in the alpha/beta (about 10-30 Hz) and gamma bands (>40 Hz). These rhythms provide insight into the most important feature of working memory, that it is under volitional control. Gamma and spiking help carry the contents of working memory, the thoughts. Beta is the control signal through which volition acts to turn on and off working memory storage.

Updating and changing theories/models is the way science works. Everything we think we know is just a stepping stone to a greater truth. 

CNS: How does such recent research change our view of working memory?

Miller: It changes our view of working memory by adding new insights. It doesn’t throw away or threaten the old model. It is an update. We all agree that spiking plays a crucial role in holding working memories. Our new model doesn’t change that.  It is a “peek under the hood” that provided new insights into how spikes help maintain working memories. New technology, new approaches, new analytics have revealed that the story is more complex than previously thought. Isn’t that always the case?

I would also add that updating and changing theories/models is the way science works.  Everything we think we know is just a stepping stone to a greater truth. People invested in current models often get defensive when you propose something new. They shouldn’t. Science should be forward-looking, not a defense of what you already think you know. I suggest reading Thomas Kuhn.

CNS: What technologies are enabling this work?

Miller: Multiple-electrode recording  Averaging data is a necessity when your focus is on individual neurons. And averaging data is a useful technique for many questions. But it obscures details of neural dynamics. Multiple-electrode recording is better suited to examining those details. It gives you more data (always a good thing) and it allows the study of interactions between simultaneously recorded neurons. You can’t really study network interactions one neuron at a time. Also important was the development of analytical tools to analyze rich multiple-electrode data.

CNS: What do you want to be the take-home message of your CNS 2019 lecture?

Miller: In addition to our new model of working memory, I guess it would be the importance of open-mindedness. Be skeptical of dogma.  Dogma is blinding. And be humble enough to be led where the data is pointing. I can think of several times in my career when new data led to a substantial revision of my way of thinking.  I am open to that because I know that I know very little about how the brain works.

CNS:  I know you also play in a band that makes regular appearances at CNS meetings. Planning to do the same for 2019?

Miller:  Pavlov’s Dogz. Your favorite cognitive neuroscientists rocking hard for you.  There will be a party/gig on Monday night [March 25]. Venue to be announced. All the cool kids will be there.

CNS: Anything else you’d like to add?

Miller: Thanks to CNS for the George A. Miller Prize. It is very gratifying to be honored by your colleagues. George was one of my teachers in grad school.  I am verklempt because I feel like we were mishpocha.

-Lisa M.P. Munoz