The parting words of Ken Jennings in last year’s Jeopardy match against Watson, a computer seemingly able to decipher and process language, are a milestone for robotic innovations. Advancements in neuroscience and robotics have focused on giving robots human-like intelligence and processing skills. This concept has been depicted numerous times in popular culture, many times in terms of robotic rebellion, for example in movies such as I, Robot or WALL-E.

Recent robotics research leaves us with a couple of questions. Are really focusing on the right aspects of advancing in robotic technologies? Instead of perfecting intelligence and processing, why not instead focus on perfecting human emotion?

Facial cues have proven extremely important for social interaction. In experiments where robots greeted humans and asked them to perform a task, the humans were more receptive when the robot glanced at the task to be performed, rather than robotically (pun intended) looking at the human subject while giving instructions. A similar experiment was set up in which human subjects were to learn about China. A map of China was present in the classroom. Those who had robot teachers who looked at the map while teaching actually learned more about the spatial relationships pertaining to the “lecture material” than those who had robot teachers who never looked at the map.

Another study examined the responsiveness of infants to robot facial cues.

Child following robot gaze

An 18-month old infant was allowed to watch a robot interact with a human (the researcher). He would point tobody parts, and the robot would repeat the action. When the researcher left the room, the infant followed the robot’s gaze. In contrast, those infants who never saw the robot interact with a human were unresponsive to their gazes. Visual communications are key for learning social interactions.

Such robots have also been used in Autism Spectrum Disorder (ASD) therapies. ASD patients have trouble with social interactions, so these social robots have been hypothesized to help in therapy. A bubble test, in which a companion to the patient blows bubbles, is used as it has been shown to provoke social interaction. ASD subjects were either allowed to interact with the robot to receive bubbles (such as by pushing a button) as well as a motor output from the robot (spinning) or could sit and watch while the robot did nothing. Those patients who were allowed to interact with the robot showed a significant increase in social behaviors such as speech and continued robot interaction. Thus, it has been concluded that the robots’ social behaviors are causing a response in ASD patients.

This work shows that robots are gaining prevalence in studying the social aspects of human intelligence. While it is still important to use robotics to study how human processing works, it will be of extreme value to also continue research in the field of emotions and social communication.

Developing Robots That Can Teach Humans – Science Nation

“Social” robots are psychological agents for infants:: A test of gaze following – Neural Networks

Toward Socially Assistive Robotics for Augmenting Interventions for Children with Autism Spectrum Disorder – Experimental Robotics

I have some news that might be a bit disappointing to…well, pretty much anyone who would find themselves on a blog dedicated to the mind and brain.  Bear with me (or not, if you’d like, really), but this is a post primarily about the heart.

I was recently introduced via a grad student in the (yes, neuroscience) lab I work in to the latest advancement in the race to perfect an artificial heart. That link is to an NPR article that really tells you everything you need to know…and you should absolutely read it.  But to summarize the details you need to know for my purposes here, the design is completely novel, and unlike previous designs, it doesn’t use nature as its inspiration.

All previous artificial hearts have attempted to mimic the beating of a natural heart, but the moving parts can wear down or cause problems such as blood clots.  Instead, this implant has only two moving rotors, spinning to move the blood continuously rather than in pulses.  Let that sink in for a second.  Yes, transplant recipients have no heartbeat.  And the first recipient lived for over a month in this state before dying of underlying problems, the “heart” still working perfectly.

So here, finally, is what all of this has to do with the brain.  The main message I took home from the NPR article and subsequent discussion (aside from, as the aforementioned grad student pointed out, the fact that one with such an implant should never accidentally fall asleep in public) is that while our instinct has previously been to imitate nature, that might not always be the most efficient logistical solution. Dr. Billy Cohn, one of the creators of the device, put it very well when he pointed out that many of the earliest attempts at flying machines had flapping wings before we realized that what works for birds and insects isn’t necessarily the best answer for us.

I decided to pass this information on to my mom, a cardiac catheterization lab RN, and someone as in love with the heart as I am with the brain.  Her eyes lit up when she realized that systole and diastole, that is, the heart’s pulsations serve no purpose aside from the maintenance of the heart itself, and the flow of blood—there’s no reason it can’t be continuous and steady if the heart itself is artificial.  “You might start seeing these soon,” I told her, “they’re the future of your field.”  “So what about you?” she replied, “How far off are artificial brains?”  I rolled my eyes at her joke.  Then it slowly occurred to me that, while still absurd for discussions of transplant purposes, just because something doesn’t function in the same way as its natural counterpart, doesn’t mean it isn’t the same thing.

So aren’t we already pretty close?

You are unique, just like everyone else.

Connectomics is the study of the structural and functional connections among brain cells; its product is the “connectome,” a detailed map of those connections. The idea is that such information will be monumental in our understanding of the healthy and diseased brain. Sebastian Seung thinks that a complete connectome of the human brain will be one of the great prizes in 21st-century neuroscience.

Efforts to construct brain connectomes are split into two categories: ones that use imaging techniques like MRI, PET, and DT, thus focusing on macroscopic connections or tracts; and those that use electron microscopy to map the tinniest of axons (0.2-20 microns in diameter) and individual synapses.

While this may sound daunting, it also seems the obvious thing to do in order to really understand how the brain works. After all, don’t all our memories, personalities, and behaviors dependent on the structure of the brain, down to the microscopic level? So why is connectomics so new? Because the three-pound enigma that can contemplate all things big and small – from protons and electrons, to planets and stars, to galaxies and the whole universe – contains more parts than anything we’ve ever studied before. The human brain, we’ve been told, holds 100 billion neurons, with close to one quadrillion synaptic connections total; storing all of that information in one brain would take one Exabyte of data (that’s one trillion Gigabytes).

Jeff Lichtman and colleages at Harvard remain hopeful. They are developing novel tools to automate the tedious task of scanning brain slices. They expect the connectome to reveal differences in the way healthy and diseased brains are wired.

The effort is laudable, considering its scope and ambition, but it begs the question: does all behavior, experience, perception, etc depend on the structure of synapses and connectivity of neurons? More pointedly, does structure determine all function – chemical and electrical? Sure, larger synapses or more dendritic spines make stronger connections and more efficient transmission of information, but a snap-shot connectome won’t take into account temporal dynamics and enzymatic processes, which play a big role in the active brain.

In his TED talk, Sebastian Seung says that to test the hypothesis that “I am my connectome,” we could try to read out memories from someone’s connectome. But memories are not just synaptic connections – they are also assemblies of neurons in time or firing sequence. The connectome does not take those into account. And Seung fails to explain how we could actually verify any of those personal memories, since current methods of constructing a connectome involve cutting the brain into thousands of 30-micron slices.

If we could devise some non-invasive methods to construct a human connectome at the synapse level, what ethical issues would we face? Could a personal connectome be the ultimate breach of privacy? Could it redefine or “undefine” what we consider to be normal brains/mental states?

Constructing a comprehensive human connectome is a great challenge. A bigger challenge would be to model the electrical dynamics of the 100 billion human neurons. But perhaps the most important quest for neuroscience isn’t building a connectome, but learning how neuronal activity creates experience.

Neurocartography – Narayanan Kasthuri and Jeff Lichtman via NIH Public Access

Sebastian Seung: I am my connectome – TED.com

Seeking the Connectome, a Mental Map, Slice by Slice – NYTimes.com