Choosing & storing permanent memories, and robot's smile | Last Week in Science (31st Mar 2024)

I too can smile

Imagine meeting a humanoid robot who can mimic your facial expressions just like humans do. Eva is one such robot. But Eva would smile only after your smile is over. Eva takes time to know what emotion you expressed. To improve this ability and make robots appear more genuine in their responses to humans, Emo has been designed by the same team of scientists at Columbia University that had designed Eva.

Emo has high definition cameras in its eyes so that it can observe the expressions of the humans it is interacting with. It has 26 actuators and magnets under its replaceable face skin so that the facial expressions can be created without missing the finer details of the human smile. 23 motors control its face and three motors its neck to achieve the required expression. Yes, Emo has better hardware than Eva and can smile much better too but what makes it appear genuine?

Videos of humans smiling, were used to train a model for Emo. After this training, Emo can now predict a human's smile 839 milliseconds before the smile comes on their face to smile with them. I don't see a smiling robot taking over the world! Do you?

Reference: Human-robot facial coexpression

Choosing long term memories to save

Our daily experiences and events are encoded and saved in the brain as memories. Some of the memories are saved permanently while others are not. We do not know how the memories are chosen by the brain for permanent storage. A recent study conducted at New York University has now provided an answer.

For this study, scientists made mice explore a maze multiple times and gave them sugary treats at the end of each run. The brain activity of the mice was recorded while they were hard at work. After some of the maze runs, a particular type of brain activity was seen called 'sharp wave ripples' in some of the neurons. This activity represents a group of neurons getting activated together repeatedly in a rhythm. For different runs of the maze i.e, for different memories, the group of neurons were also different.

When the mice slept after all these maze runs and recordings, the rhythmic activity of neurons for some of the memories were repeated 1000s of times. The brain replayed only those memories that had shown wave ripples 15 - 20 times during the maze runs and others were not replayed. The replaying of memories 1000s of times strengthens the memory to store it for long term. Thus, the brain uses sharp wave ripple activity to select a memory for permanent storage.

Reference: Mechanism found to determine which memories last

How do memories get stored permanently?

According to our current understanding, our memories are encoded in micro circuits of neurons that form in the hippocampus. A micro circuit of neurons is called an engram and it represents the physical form of a memory. For the engram to form and persist, the individual neurons of the engram undergo biochemical changes to maintain strong connections with other neurons of the engram. One such change is damage to the neuron's DNA that gets corrected immediately.

A new study has found that there are neurons other than engram neurons in the hippocampus in which DNA breaks in response to/for memory formation but do not get repaired immediately. These neurons activate a protein called TLR9 in response to the DNA damage that further activates other proteins which help to repair the DNA damage. When scientists removed TLR9 from mice and then tested their memory, they found that the mice could not remember what they learnt. 

This study also shows that there are other group of neurons in addition to the engram neurons that are responsible for storage of permanent memories.

DNA damage is usually associated with diseases like cancer, neuro psychiatric and neurological conditions like neurodegeneration, where memory and learning is often affected. It is therefore intriguing that how such a dangerous step like DNA damage is used by our neurons for storing memory. A better understanding of this mechanism of memory storage would help in saving neurons, their connections and micro circuits, in conditions where we start to lose them.

Inputs & editing by Ashish Gourav