Correction appended.
When we look at an object like a coffee cup, does it get recognized by a single cell in the brain? By 20? By 20,000? Would the same neurons fire if we were to look at a different object, like an ice cream cone?
In a study published Thursday in the journal Neuron, Professor of Neuroscience David Sheinberg and Luke Wolosyzn GS argue for the existence of small neural networks that fire up when we look at different images.
For decades, researchers have studied and debated the Grandmother Cell hypothesis - the idea that a few cells might be responsible for recognizing highly specific images, such as the face of your grandmother. "We don't know if the way you see something complex in the real world is based on activation of just a small number of cells or actually a distributed pattern of activity spread across all 30 areas and millions of neurons," said Professor of Neuroscience Michael Paradiso, who was not involved in the study.
The brain contains many classes of neurons, which play different roles in object recognition, Sheinberg said. Excitatory pyramidal neurons may be the only cells to recognize specific objects, while other neurons, such as inhibitory interneurons, respond more to unfamiliar objects. Previous studies averaged together the contributions of many classes of neuron, which obscured the impact of specialized image cells, he added.
Many researchers have looked for specialized neurons and failed to find them, said David Freedman, assistant professor of neurobiology at the University of Chicago, in a response to the study, also published in Neuron. The strength of the study, Freedman said, is that it identifies how and why previous research went wrong and demonstrates new methods for locating specialized image cells. Previous researchers did not expose subjects to a sufficiently wide spectrum of images, the study argues. This study exposed subjects to a gallery of 300 pictures, including a flashlight, a dartboard, a hand of playing cards and an old wooden chair.
But Sheinberg said there are some problems with the Grandmother Cell hypothesis, though his findings could support the hypothesis. "There's not a lot of redundancy" of cells in the brain, Sheinberg said. "If there are truly a few, and you lose them, then are you saying it's gone? That seems absurd."
"Specificity isn't always a good thing," Sheinberg added. "If you want to recognize your mom, you want to recognize her in a somewhat invariant way. You don't care if she's wearing a red sweater."
But speedy recognition of specific objects might benefit us at times, Sheinberg said. Imagine, for example, a ball flying toward your face - wouldn't you want to see it quickly?
The study concludes that responses to familiar objects become stronger over time as the brain becomes more practiced in recognizing objects. The neurons that fire together wire together, resulting in stronger connections between those cells.
Sheinberg's study is part of the REPAIR project, an army-funded initiative to research and develop brain implants that could aid patients with traumatic brain injuries. One of Sheinberg's hypotheses is that stimulation of inhibitory interneurons might send the brain "a learn signal," leading to more effective recognition of objects. The current study could lead to brain implants that stimulate inhibitory interneurons, helping treat conditions like prosopagnosia, an inability to recognize faces that can result from brain injury.
"Our goal is to try to understand the circuits enough to say, 'If part of it breaks, are we going to be able to insert artificial signals?'" Sheinberg said.
Examining monkey brains could be a good model for understanding human behavior, since monkey and human brains are quite similar, Paradiso said.
A previous version of this article included a quote characterizing the monkeys in the lab. The quote was removed from the article because, out of context, it did not accurately portray the source's views on the treatment of animals in labs.
