Jul 06 2021

The Grandmother Neuron

People are really good at recognizing faces. One of my lecture bits is to show a picture of a famous person for a fraction of a second (as fast as I can make the picture cycle in Powerpoint). The vast majority of the audience has no problem placing this face in their memory from just a quick glance. How does the brain accomplish this feat? That is something that neuroscientists are still working out. We already have a lot of information about the visual cortex and how it puts images together. We also know that the fusiform gyrus is essential for facial recognition. This brain structure lies at the junction of the occipital lobe, where the visual cortex is, and the medial temporal lobe which is involved with memory.

What is not well understood is exactly how the visual cortex connects to memory – how do we recognize familiar faces? One hypothesis is somewhat dismissively known as “the grandmother neuron”, based on the idea that one neuron would encode the face of your grandmother and connect that pattern with your memories of your grandmother. No such neuron has been discovered, and advances in neuroscience have led to the conclusion that it likely never will be. One neuron is insufficient to encode something as complex as a human face.

But perhaps there is a region or a circuit or network in the brain that serves this function. Neuroscientists may have zeroed in on the brain region that functions like the grandmother neuron. They studied macaque monkeys, exposing them to pictures of faces they have seen in person and other they have not seen before except digitally. When pictures of faces they have seen before were displayed the temporal pole (front tip of the temporal lobes) lit up on fMRI scanning. This cluster of cells also responded “non linearly” when increasing amounts of a familiar faces were shown, implying that the whole pattern is greater than the sum of its parts. Further, when distinctive parts of a face were show they lit up as if the entire face were show. This could indicate that once sufficient detail were shown to trigger the memory of the entire face, full activity of that region was then engaged.

Obviously there is much more to study here. No one expects teasing apart how something as complex as the brain works to be easy. But this does appear to be an important piece to the puzzle. What is likely going on is that this cluster of neurons are together acting as a fast connection between sensory input and memory. Rapid facial recognition is evolutionarily important, so it makes sense that there would be dedicated brain areas to make this happen fast. The fusiform gyrus is one area, dedicated to facial patterns. This is partly why primates and especially humans are so partial to the pattern of a face. We see it in the clouds and the random patterns that make up the barks of trees, oil stains, and shadows on Mars (in a process known as pareidolia).

The authors of the current study suspect that the neurons they are looking at in the temporal pole are hybrid neurons – involved in both visual sensation and memory. They may be an actual new kind of neuron (there are many neuron types with different configurations of axons and dendrites to serve their specific purpose). What may be happening is that these neurons serve as a juncture between the part of the visual cortex that puts together images and that part that remembers known faces. When a match is found, it lights up. There must be other steps involved in this process, but the authors seem to have found a critical juncture.

Something like the following is likely happening – when you see any image and the brain construct from a pattern of light, color, and shadow into an actual object, that pattern is compared in real time to your visual memory. If the image is obvious, then the process is seamless. If it is ambiguous, you may experience your brain trying to find a match – “what is that?” When the brain finds a probable match, then the image snaps into clarity. This is partly because your brain modifies the image to make it look more like the apparent match. With faces the same thing happens but uses dedicated brain areas for just faces, so that the process is faster and more discriminating.

Think of all those times you saw an actor on a show that you know you recognize, but you can’t place. In the background your brain is searching your catalogue until it finds a match. Or, you may not recognize the actor (because of makeup and acting obscuring familiar details) until suddenly you see them from a particular angle or they may a familiar gesture or expression, then your memory is triggered. Also, you may recognize a familiar face just from their eyes or their mouth. That part is enough to trigger the memory of the whole. But there is some threshold where recognition becomes complete, like a switch being flipped.

Neuroscientists are now one step closer to reverse engineering how the brain produces these phenomena. The key component is a connection between perception and memory, and for facial recognition this study provides evidence that the temporal pole may be a critical pieces of this connection.


No responses yet