The study is based on brain recordings from patients with intractable epilepsy, who had electrodes implanted deep in their hippocampus to identify where their seizures were occurring.
These implants also give scientists the opportunity to study how individual neurons in the hippocampus fire during wakeful activity.
When a participant was shown a picture of Shrek, researchers noticed a particular neuron would fire in their hippocampus.
This same neuron also activated when volunteers were asked to read a passage about Shrek and Courtney Love and the name 'Shrek' appeared.
Gradually, the activity of that neuron began to fade, but when the pronoun "he" was used in a later sentence to refer to Shrek, the same 'Shrek' neuron activated. When volunteers read the pronoun "she", on the other hand, it did not activate the same cell.
What's more, if Shrek was absent from the first sentence, the pronoun 'he' did not activate the 'Shrek' concept cell. This suggests that only pronouns assumed to be referring to Shrek activate the concept neuron.
I'm mostly amazed by the techniques deployed to draw these kind of conclusions.
Published in Science.