Minds, brains and woo.

Mind is not a redundant concept.  I'm with Andrew Brown in this short piece from the Guardian (17 June 2011), 'Mind, brains and woo.'  Here's the ending:

There is something very odd about the idea that the mind is an illusion that a brain has about itself (which is what is implied in a lot of this talk). Illusions are themselves things that only minds can have. An illusion, or a delusion, demands that there is a subjectivity being deluded. If a Buddhist says that the world is an illusion, at least they are being consistent, in that they suppose the ultimate reality is more like a mind – the kind of thing that can have an illusion or can be deluded. But no one can fool a rock, or a computer. Why should a brain be different? 

Modeling simple worms.

302 neurons - It's a start!

'Is This Virtual Worm the First Sign of the Singularity' by Alexis Madrigal at the Atlantic May 17, 2013 (poor title, but good article) describes efforts to computationally model relatively simple life form, the C. elegans worm.  The project raises interesting questions about a number of things, including what is life? and what constitutes understanding?

"If you're going to understand a nervous system or, more humbly, how a neural circuit works, you can look at it and stick electrodes in it and find out what kind of receptor or transmitter it has," said John White, who built the first map of C. elegans's neural anatomy, and recently started contributing to the project. "But until you can quantify and put the whole thing into a computer and simulate it and show your computer model can behave in the same way as the real one, I don't think you can say you understand it."

This species of worm apparently has 959 cells, 302 of which are neurons.  These neurons form approximately 10,000 connections.  Part of what is unknown is just how much of the behavior of each cell needs to be fully modeled in order to simulate the worm's behavior.  I think it's probably a pretty good place to start.

I asked several researchers whether simulating the worm was possible.  "It's really a difficult thing to say whether it's possible," said Steven Cook, a graduate student at Yale who has worked on C. elegans connectomics. But, he admitted, "I'm optimistic that if we're starting with 302 neurons and 10,000 synapses we'll be able to understand its behavior from a modeling perspective." And, in any case, "If we can't model a worm, I don't know how we can model a human, monkey, or cat brain."

Brain Stimulation - a net positive?

 'The Hidden Costs of Cognitive Enhancement' from Greg Miller, Wired (March 5, 2013), points to some findings that show there are tradeoffs.  There are a number of studies going on using low levels of electrical stimulation to particular areas of the brain while attempting cognitive tasks such as using a new abstract math system.  While there appear to be some findings for faster learning, researchers Cohen Kadosh and his colleague Teresa Iuculano at University of Oxford also looked for follow-on effects.

Those who had the parietal area involved in numerical cognition stimulated learned the new number system more quickly than those who got sham stimulation, the researchers report today in the Journal of Neuroscience. But at the end of the weeklong study their reaction times were slower when they had to put their newfound knowledge to use to solve a new task that they hadn’t seen during the training sessions. ”They had trouble accessing what they’d learned,” Cohen Kadosh said.

See also this story 'Electrical Brain Stimulation Helps People Learn Math Faster' - also by Miller - May 16, 2013.

Who's in charge?

Gazzaniga's 2011 book reviews the neuroscientific evidence.

Who's in Charge? by Michael Gazzaniga (he's famous for the early split brain studies, finding some of the ways the two hemispheres differ) is a pretty light read on current neuroscience, with the angle of looking at the idea of free will and what it means for personal responsibility and law.  He covers some of his own findings, in particular around the left brain 'interpreter' which appears to be very good at making up stories to fit the apparent evidence.

While Gazzaniga lays out the strict reductionist/determinist viewpoint very effectively, he backs away from that outlook, and describes interlocking, complementary systems of upward and downward causality.  He quotes a computer analogy from David Krakauer that makes good sense to me:

We do not program at the level of electrons, Micro B, but at a level of a higher effective theory, Macro A (for example, Lisp programming) that is then compiled down, without loss of information, into the microscopic physics. Thus, A causes B. Of course, A is physically made from B, and all the steps of the compilation are just B with B physics. But from our perspective, we can view some collective B behavior in terms of A processes. (p. 139)

Moving over to the world of the brain, he takes a shot at the importance put on Libet's findings of brain activity preceding conscious awareness of movement:

What difference does it make if brain activity goes on before we are consciously aware of something? Consciousness is its own abstraction on its own time scale and that time scale is current with respect to it. Thus, Libet's thinking is not correct. That is not where the action is, any more than a transistor is where the software action is. (p. 141)

He goes on to investigate the idea of the social mind - how our behavior is shaped and constrained by the actions of others, some directly and some more indirect via cultural norms.  This is a whole other emergent level of causation that the reductionist viewpoint cannot really describe.

The last portion is about the legal aspects, and here I was a bit surprised that he did not reference David Eagleman since that his a big area of his focus, and overall it felt a bit too light a review to really do it justice.

But overall I enjoyed the book, agreed in general with his scientific and philosophical take on things.

Here's a review from WSJ online by Raymond Tallis.

World Wide Mind - is it really coming? Maybe so.

Michael Chorost's 2011 book World Wide Mind makes the case for the 'coming integration of humanity, machines and the internet'.  Taking off from Rebuilt, the book predicts that over the next 30 years or so many people will use direct brain implants to both receive inputs from others and broadcast out meaningful impulses in some fashion.

Overall I felt the science here is pretty believable.  We are learning very quickly about the brain, and how to detect neuron groups (cliques) that correlate with certain concepts.  One can imagine this work continuing at a rapid pace.  Chorost's experience with cochlear implants lends credibility to the technical progress in the area of direct stimulation of neurons to produce valuable sensory input.  He describes some nano-technology possibilities for putting outside tech in touch with many areas of the brain (more invasive in terms of its reach, less invasive in terms of not requiring head surgery).

The picture he paints in the book is one of constant reception of low-level inputs from other people you are linked to, for instance if a co-worker has some new ideas, you could receive inputs about the cluster of concepts connected to the idea, and along with standard written communication this could help trigger new ideas in you.  Chorost rightly points out that these inputs are not going to give you the other person's experience - rather you will get some inputs that will trigger your own experience given your own memories, etc.

There are two main points that I think deserved more attention in this book.  The first is the why question.  Why will people feel compelled to have these types of inputs?  One example from the book did not make this case well at all:

Having brainlike computers would greatly simplify the process of extracting information from one brain and sending it to another.  Suppose you have such a computer, and you're connected with another person via the World Wide Mind.  At the moment you're observing each other's visual experiences.  You see a cat on the sidewalk in front of you.  Your rig is able to watch neural activity in your neocortex with its optogenetic circuitry.  It sees activity in a large percentage of neurons constituting your brain's representation of a cat.  To let your friend know you're seeing a cat, it sends three letters of information - CAT - to the other person's implanted rig.  That person's rig activates her brain's invariant representation of a a cat, and she sees it. (p. 135)

Wow - distinctly underwhelming!  All this tech to send a three letter message, and wouldn't the receiver have the same experience if they received a text message with the word 'cat'?  To be fair, Chorost does give somewhat more compelling examples later in the book, but after this weak start I was pretty skeptical.

The second issue - let's say everyone had such technology implanted.  It seems to me that many commercial forces would be chomping at the bit for the ability to send inputs to everyone constantly - fast food joints might want to send you ideas of hamburgers and milk shakes, and so on.  And what kind of inputs might a government wish to send out to citizens?  Chorost spends a couple pages (196-8) on these types of questions in the final chapter, but for me it was a case of too little too late.

Overall an interesting book, one that imagines some potential developments in the field, is perhaps a bit optimistic about our ability to solve each hard problem that arises, and is sincere in its investigations of how such technology could connect us in new, meaningful ways.  The book also includes some touchy-feely material about a workshop the author attended, and it may turn off some readers.

Here's a link to the NY Times review by Katherine Bouton.  Here's another more critical blog reaction from Backreaction (two physicists).

Wagging a rat's tail via EEG signal - not much here.

This week saw many stories on the experiment done by Seung-Schik Yoo of Harvard Medical School.

"Interspecies telepathy: human thoughts make rat move" by Sara Reardon in NewScientist sums it up pretty well (bad title by the way, I'd say, since wires were involved!).

The human volunteers wore electrode caps that monitored their brain activity using electroencephalography (EEG). Meanwhile, an anaesthetised rat was hooked up to a device that made the creature's neurons fire whenever it delivered an ultrasonic pulse to the rat's motor cortex.

But the trigger detected by the EEG was simply a change in the person's concentration, which was enough to send the pulse. As Ricardo Chavarriaga (of the Swiss Federal Institute of Technology) comments:

More importantly, Chavarriaga says, the experiment will not be meaningful until the human's intention corresponds with the rat's action. For instance, a person might imagine moving their left hand to move the rat's left paw. Yoo's approach would not be of any use for that because it only tells us that a person's mental focus has changed, not what the thought or sensation behind the change is.

So overall I think this is not so interesting, just one more fairly minor step.

Michael Chorost's Rebuilt (2005).

Tells of regaining hearing via cochlear implant - 16 pins that feed impulses into Chorost's neurons near the ear.  Those of us with organic hearing are using a bunch of tiny hairs to pick up sound waves, and we likely don't think too much about how it all works, or whether we are really getting a 'realistic' sonic landscape.  But Michael Chorost suddenly lost his hearing in 2001, and chose to get the surgery to have an implant placed in his head, and Rebuilt tells the story in a compelling fashion by mixing the science, his experience of hearing, the concept of the cyborg, and his thoughts on communicating and connecting with other people.

While I've read a number of things that describe the neuroscience of vision in some detail, this was the first book I've found that does something similar for the auditory sense.  The full rig that makes the implant useful includes a microphone, a processor that can run software to scan the sound and adjust/filter the input to create output for the 16 pins, and a radio relay unit that is magnetically linked to the implant on the outside of the skin on the skull.  Many rounds of mapping are performed to fine tune the processing to the specifics of the way the pins transmit data into the brain, and new software in the processor can make improvements by (for example) increasing the transmission rate.

One of the stories that really brought the whole concept home for me was telling of hooking up the rig to a CD player, such that no sound waves were produced, simply electronic patterns, which were transmitted to the implant and created the experience of hearing for Chorost.  Just as our vision is a creation of the brain, so too our auditory sense.  And it's quite interesting to track how his hearing ability improves over time, both due to software enhancement and neuro-plasticity of the brain.

Chorost mulls over the concept of the cyborg quite extensively through the book, contrasting this concept of the technologically enhanced human with the quite different notion of a robot.  Knowing that your hearing is dependent on the hardware and software running on various gizmos will do that to a person.  He stresses the point that his implant does not improve on healthy human hearing, and takes issue with some of the more extravagant claims of Kevin Warwick (who wrote in I, Cyborg, "We will interface with machines through thought signals.")  We don't really have any clue yet how thoughts might be represented as an interface to the brain - in this situation what's being transmitted are representations of sound, not thought!

The brains of rats...

Go better together!  Curious story today in the Guardian, "Brains of rats connected to allow them to share information via internet" by Ian Sample (Feb 28, 2013).  The internet part of it seems pretty gimmicky, not sure why that's important, and it's a little unclear to me what is really going on here.  The substantive portion is described like this:

The scientists first demonstrated that rats can share, and act on, each other's sensory information by electrically connecting their brains via tiny grids of electrodes that reach into the motor cortex, the brain region that processes movement.

The rats were trained to press a lever when a light went on above it. When they performed the task correctly, they got a drink of water. To test the animals' ability to share brain information, they put the rats in two separate compartments. Only one compartment had a light that came on above the lever. When the rat pressed the lever, an electronic version of its brain activity was sent directly to the other rat's brain. In trials, the second rat responded correctly to the imported brain signals 70% of the time by pressing the lever.

Remarkably, the communication between the rats was two-way. If the receiving rat failed at the task, the first rat was not rewarded with a drink, and appeared to change its behaviour to make the task easier for its partner.

So it sounds like impulses from one rat, which were generated upon a certain movement, were sent in some format into the other rat's brain, into its motor cortex.  That in turn seemed to influence the movements/'decisions' of the receiving rat.

Some questions I have about this - (1) how much of a learning period was involved? 2) how many notable movement/decision options did the receiving rat actually have?  3) was there some timing boundary within which the rat had to take action to be seen as successful communication? 4) what was the success rate? Maybe it's all answered in the paper.

For me it raises the question of whether essentially any patterned impulses received by certain brain areas could be successfully 'interpreted' - and to what level of discrimination.  In this case the receiver may be doing little more than using the reception as a timing signal to make a movement, but perhaps it can go deeper than that.

The article ends with a good reminder that there's plenty we don't know in this area:

Very little is known about how thoughts are encoded and how they might be transmitted into another person's brain – so that is not a realistic prospect any time soon. And much of what is in our minds is what Sandberg calls a "draft" of what we might do. "Often, we don't want to reveal those drafts, that would be embarrassing and confusing. And a lot of those drafts are changed before we act. Most of the time I think we'd be very thankful not to be in someone else's head."

(H/t to twitterers @pourmecoffee and @neurophilosophy)

Update:  Here's another report from NYT:  "One Rat Thinks, and Another Reacts"