Our brains are brilliant at separating the picture from our eyes into objects, remembering them, and comparing them to similar objects we have seen before. Every day, we take this for granted – imagine how long it would take to leave the house each morning if we had to constantly keep learning what each household object is, whose it is, and what its used for. Just how our brain does this, though, has been debated for years.
Perceiving our visual surroundings requires the work of a complex web of areas at the back of the brain, whose functions are arranged in a hierarchy from basic perception of light and shade, up to the grouping of shapes into a coherent object. Further up the hierarchy, neurons are pickier about what they respond to. For instance, an area coined the ‘fusiform face area’ appears to respond specifically to faces, or shapes arranged to look like faces. But in these more refined visual areas, it is not clear exactly how this sophisticated recognition occurs. In the extreme is the idea that, at the very top of the visual hierarchy, there may be cells responding not just to people, but a single individual. This quirky idea became known as the Grandmother Cell Theory, as a flippant comment by neuroscientist Jerry Lettvin stuck.
Although this idea is not widely accepted, one team of perception researchers were astonished to find something remarkably similar: a neuron that seemed to respond to pictures of Jennifer Aniston, and nothing else.
The team were investigating the activity of individual neurons in eight patients with treatment-resistant epilepsy. They had previously had several depth electrodes fitted as part of their Epilepsy treatment, to find the source of their seizures. Fitting electrodes deep in a human brain is a fairly invasive and risky procedure, and would be ethically dubious unless part of a beneficial treatment. Ergo, epilepsy patients make an ideal sample for researchers needing to use electrodes: as technology stands at the moment, you can’t record from individual neurons any other way. The areas recorded from lie at the junction between long-term memory and the higher stages of visual processing, in the medial temporal lobe. From previous research, the team had noticed cells in these areas which consistently responded to particular people, so they decided to investigate further.
The epilepsy patients were shown still images of many different people and landmarks, one at a time, whilst neural activity was monitored. Pictures eliciting a response, which differed between patients, were retained. Alternative images of the same people or objects were then shown to the patients. When a neuron did respond to a picture, it was usually highly selective, only reacting strongly to a particular person or landmark. For instance, a cell in the left hippocampus of one patient responded to pictures of Jennifer Aniston, but gave no meaningful reaction to any other people, animals or landmarks.
On closer inspection, it appeared that the ‘Jen-ron’ was responding not to Jen’s physical features, but to the abstract concept of her. The Jen-ron also responded to pictures of Lisa Kudrow, her Friends co-star, but, interestingly, not to pictures of her ex-husband, Brad Pitt. Similarly, another neuron responded preferentially to pictures of Halle Berry, but also to a drawing of her, her name written down, and a picture of her as Catwoman. In the latter picture, Halle’s face was hidden, but he owner of the neuron was aware that she had played Catwoman. Visual clues didn’t seem necessary to unlock the concept of Halle Berry and evoke a reaction from the cell.
Though the findings seemed robust, some experiments haven’t found the same results, and there are plenty of caveats. How many Jen-rons can a person possibly have – enough for everyone we know? Do they only code for one person? After all, the researchers only used a limited number of pictures – it is plausible that these cells responded to other people or objects not used in the experiment. Do we only have them for people who are important to us, or whom we see a lot? How much exposure to a person is necessary for us to develop a Jen-ron for them? Do they expire if we don’t see that person for a while?
The lead author of the study, Quian Quiroga, is keen to stress that these are not Grandmother cells. Instead, the paper concludes that this is evidence for ‘sparseness,’ the notion that individuals or objects can be recognised by a relatively small number of neurons that respond to very specific criteria, rather than a broad network where neurons work more closely together, each responding to different basic features, all at the same time, to achieve recognition. Most researchers think the answer lies somewhere between two ideas. After all, some of the interpretation must happen outside the Jen-ron – it is highly improbable that, should one of these cells die, the person it responded to would be entirely erased from memory.
No neuron is an island, so it is important to investigate how these neurons interact with visual and memory areas – they will still depend on inputs from simpler visual processing areas and are likely to interact with memory. However contentious though, Jennifer Aniston cells, whether or not they exist, open wide the question of how the brain ascribes meaning to entities.