Such cells, together with their sisters—for there are probably thousands of such cells in the medial temporal lobe for any one idea—encode a concept, such as Jennifer Aniston, no matter whether the patient sees or hears her name or looks at her picture. Think of them as the cellular substrate of the Platonic ideal of Jennifer Aniston. Whether the actress is sitting or running, whether her hair is up or down, as long as the patient recognizes Jennifer Aniston, those neurons are active.
Nobody is born with cells selective for Jennifer Aniston. Like a sculptor patiently releasing a Venus de Milo or Pietà out of blocks of marble, the learning algorithms of the brain sculpt the synaptic fields in which concept neurons are embedded. Every time you encounter a particular person or object, a similar pattern of spiking neurons is generated in higher-order cortical regions. The networks in the medial temporal lobe recognize such repeating patterns and dedicate specific neurons to them. You have concept neurons that encode family members, pets, friends, co-workers, the politicians you watch on TV, your laptop, that painting you adore.
Conversely, you do not have concept cells for things you rarely encounter, such as the barista who just handed you a nonfat chai latte tea. If you were to befriend her, meet her later in a bar and let her into your life, the networks in the medial temporal lobe would recognize that the same pattern of spikes occurred repeatedly and would wire up concept cells to represent her.
Concept cells demonstrate compellingly that the specificity of conscious experience has a direct counterpart at the cellular level. Say you are recalling the iconic scene of Marilyn Monroe standing on a subway grill, trying to keep the wind from blowing her skirt up. This conscious percept will be caused by a coalition of neurons numbering perhaps in the hundreds or thousands rather than in the billions, as is commonly assumed.
Making Concepts Visible
More recently, Moran Cerf and others from my lab, together with Fried, hooked several concept cells to an external display to visualize a patient’s thoughts. The idea is deceptively simple but fiendishly difficult to implement. It required three years of effort by Cerf, a computer-security specialist and a moviemaker turned Caltech graduate student, to pull off this feat. Let me walk you through one example. Cerf recorded from a neuron that fired in response to images of actor Josh Brolin (whom the patient knew from her favorite movie, The Goonies) and from another neuron that fired in response to the Marilyn Monroe scene I just mentioned. The patient looked at a monitor where these two images were superimposed, with the activity of the two cells controlling the extent to which she saw Brolin or Monroe in the hybrid image.
Whenever the patient focused her thoughts on Brolin, the associated neuron fired more strongly. Cerf arranged the feedback such that the more this cell fired relative to the other one, the more visible Brolin became and the more the image of Monroe faded, and vice versa. The image on the screen kept changing until only Brolin or only Monroe remained visible and the trial was over. The patient loved it, as she felt that she controlled the movie purely with her thoughts. When she focused on Monroe, the associated neurons increased their firing rate, the cells for the competing concept, Brolin, dampened their activity, whereas the vast majority of neurons remained unaffected.
It might appear as if there are two people involved in this experiment, the way the puppeteer Craig occupied the head of actor John Malkovich in the 1999 movie Being John Malkovich. One is the patient’s mind, instructing her brain to think of Monroe. The other is the one that is acting out the mind’s desire—namely, the nerve cells in the medial temporal lobe that up- and down-regulate their activity accordingly. But both are part of the same brain. So who is in control of whom? Who is the puppeteer, and who the puppet?