How much neuroscience in 'Social'?

Psychologist Matthew Lieberman does like the fMRI!  In his new book 'Social' (2013) the UCLA professor and Director of the Social Cognitive Neuroscience Lab makes the case for the neural underpinnings of our social learning and behavior.  The question that came to my mind though was how much of the message was basically social psychology (which is valuable, don't get me wrong, but not dependent on fMRI findings).

The book features many diagrams of brains, pointing out various regions that are active during different cognitive tasks.  In general the correlations of active areas to cognitive tasks can be very useful to better understand the brain structures, if not to actually understand how the cognitive tasks are achieved. Most illuminating are the findings where either the same area is used during different types of tasks, or where different areas are used for what seem to be very similar tasks.  I think it's probably valuable to combine these types of findings with traditional psychology to see what may be illuminated.

Lieberman's key claim is that our 'default' brain mode is used for so-called 'mentalizing' - sorting through the social world, trying to understand other people's motives and intentions. This is shown by the activation of certain brain areas both while explicitly thinking about social problems and when not attempting to do other cognitive tasks.

We typically use a particular prefrontal brain region for general cognition (reading, memorizing, computing, etc.), and it was thought that these areas were the critical to all learning.  But various studies have found a 'social encoding advantage' in learning using the mentalizing system to form overall impressions of people and their intentions rather than simple memorization of people's behavior.  The finding was that 'the folks making sense of the information socially have done better on memory tests than the folks intentionally memorizing the material.' (284)  From the neuroscience angle:

Jason Mitchell, a social neuroscientist at Harvard University, ran an fMRI version of the social encoding advantage study. As in a dozen studies before his, he found that when people were asked to memorize the information, activity in the lateral prefrontal cortex and the medial temporal lobe predicted successful remembering of that information later on. According to the standard explanation of the social encoding advantage, the same pattern should have been present or event enhanced when people did the social encoding task, but that isn't what happened. The traditional learning network wasn't sensitive to effective social encoding. Instead the central node of the mentalizing network, the dorsomedial prefrontal cortex, was associated with successful learning during social encoding. (284-5)

Lieberman suggests a number of interesting applications of this finding to change and hopefully improve the way we teach kids, who are intensely interested in the social world and not so interested in memorizing facts - such as by teaching history more in terms of the social dramas (rather than actions and dates), and math by engaging students as both tutors and tutees.

The book has sections on three stages of social development, which he terms connection, mindreading (theory of mind), and harmonizing - and argues that significant brain resources are devoted to maintaining connection with other people.  Harmonizing is about taking on many of the goals and behaviors of our social group (particularly active during adolescence).  The idea here is that our sense of self as supported in the brain is very susceptible to the social messages we receive.

Overall I liked this book - not that it really lives up to the subtitle 'Why Our Brains Are Wired to Connect' - it's more about 'How' than 'Why'. At its best it reminds us that we are truly social creatures, and the neuroscience helps illustrate that point.

Will we understand science in the future?

Tyler Cowen suggests not in his book 'Average Is Over' (2013).  The book is a bit of prognostication about the near future, looking mainly at how the use of computers is and will change our world.  The basic idea is that the people who can add value to computer work in some way will reap most of the rewards.

For the purposes of this blog, I thought the part about computer-driven science was most interesting. Cowen lists three reasons why science may become harder to understand:

1. In some (not all) scientific areas, problems are becoming more complex and unsusceptible to simple, intuitive, big breakthroughs.
2. The individual scientific contribution is becoming more specialized, a trend that has been running for centuries and is unlikely to stop.
3. One day soon, intelligent machines will become formidable researchers in their own right. (206)

And here's one attempt at a summary:

The remaining human knowledge of science will be very practical, very prediction-oriented, and well geared for improving our lives.  Of course those are all positive developments. Still, as a general worldview, science will not always be very inspiring or illuminating. The general educated public will to some extent be shut out from a scientific understanding of the world, and we will run the risk that they might detach from a long-term loyalty to scientific reasoning. (219)

It will be interesting to see how much of this thinking will apply to neuroscience.

Brain decoding - how far can it go?

Kerri Smith has a good overview of the topic in "Brain decoding: Reading minds" at Nature.  The range of investigation goes from identifying the content of dreams to verifying whether someone is lying, to trying to understand the full process of how the brain can encode information.  But the starting point is fairly modest - trying to identify what object someone is looking at based on patterns in the visual area of the brain.  There's a good reason to start there:

Applying their techniques beyond the encoding of pictures and movies will require a vast leap in complexity. "I don't do vision because it's the most interesting part of the brain," says Gallant. "I do it because it's the easiest part of the brain. It's the part of the brain I have a hope of solving before I'm dead." But in theory, he says, "you can do basically anything with this."

But of course theory and practice are two different things, and there may be practical limits:

Devising a decoding model that can generalize across brains, and even for the same brain across time, is a complex problem. Decoders are generally built on individual brains, unless they're computing something relatively simple such as a binary choice — whether someone was looking at picture A or B. But several groups are now working on building one-size-fits-all models. "Everyone's brain is a little bit different," says Haxby, who is leading one such effort. At the moment, he says, "you just can't line up these patterns of activity well enough."

Using this kind of research to detect 'secret' product preferences seems pretty misguided to me.  But that doesn't stop some from trying!

Decide what you think - it matters!

Tom Stafford at mindhacks.com writes on free will studies that indicate some interesting side effects of reading about a deterministic model.  Here's the bottom line:

This is a young research area. We still need to check that individual results hold up, but taken all together these studies show that our belief in free will isn’t just a philosophical abstraction. We are less likely to behave ethically and kindly if our belief in free will is diminished.

Personally I do think that regardless of the exact underlying physical mechanisms, one's choices help set the pattern for future behaviors, so best to act carefully and with fore-thought!

Follow-up on the brain-to-brain experiments

Mind Hacks blog presents a nice short analysis of the UW experiment ("It is mind control but not as we know it"), written by Tom Stafford.  Previously I logged an entry for the brain-to-brain communication experiment conducted at University of Washington by Rajesh Rao.  Here's the gist from Stafford:

In information terms, this is close to as simple as it gets. Even producing a signal which said what to fire at, as well as when to fire, would be a step change in complexity and wasn’t attempted by the group. TMS is a pretty crude device. Even if the signal the device received was more complex, it wouldn’t be able to make you perform complex, fluid movements, such as those required to track a moving object, tie your shoelaces or pluck a guitar. But this is a real example of brain to brain communication.

As the field develops the thing to watch is not whether this kind of communication can be done (we would have predicted it could be), but exactly how much information is contained in the communication.

Human-to-human brain communication

A very limited form of brain-to-brain communication is described in a story on research at the University of Washington: "Researcher controls colleague’s motions in 1st human brain-to-brain interface" by Doree Armstrong and Michelle Ma, Aug 27, 2013. The experiment used EEG signals via Skype to transmit signals of thoughts of simple movement, which the receiver got via transcranial magnetic stimulation - "a noninvasive way of delivering stimulation to the brain to elicit a response.... in this case, it was placed directly over the brain region that controls a person’s right hand."

I believe there are quite severe limits to the type of signal which could actually be transmitted and received via this mechanism, and the researchers confirm:

At first blush, this breakthrough brings to mind all kinds of science fiction scenarios. Stocco jokingly referred to it as a “Vulcan mind meld.” But Rao cautioned this technology only reads certain kinds of simple brain signals, not a person’s thoughts. And it doesn’t give anyone the ability to control your actions against your will.

Both researchers were in the lab wearing highly specialized equipment and under ideal conditions. They also had to obtain and follow a stringent set of international human-subject testing rules to conduct the demonstration.

“I think some people will be unnerved by this because they will overestimate the technology,” Prat said. “There’s no possible way the technology that we have could be used on a person unknowingly or without their willing participation.”

Thin slicing the brain.

Creates a whole lot of data!  Nature reports on 'Whole human brain mapped in 3D' by Helen Shen, June 20, 2013.  The atlas was created from 7400 slices of a human brain, each thinner than a human hair, and nicknamed 'BigBrain'.  Here's the quick summary:

The brain is comprised of a heterogeneous network of neurons of different sizes and with shapes that vary from triangular to round, packed more or less tightly in different areas. BigBrain reveals variations in neuronal distribution in the layers of the cerebral cortex and across brain regions — differences that are thought to relate to distinct functional units.

Given that we are still working on a model for a simple worm with 302 neurons, there's obviously a long way to go with the full human brain.  But you gotta start somewhere, and I'm sure that having an accurate map will help (now just drawn from one example, but as they do more they will get an idea of the individual differences that are possible - I'll bet they can be pretty significant).

What's the program in the Chinese Room?

It's really big and complicated!  That's my main takeaway from the Dennett writing on the Searle thought experiment (in Intuition Pumps and other books).

Here's the description of the scenario from Wikipedia:

It supposes that there is a program that gives a computer the ability to carry on an intelligent conversation in written Chinese. If the program is given to someone who speaks only English to execute the instructions of the program by hand, then in theory, the English speaker would also be able to carry on a conversation in written Chinese. However, the English speaker would not be able to understand the conversation. Similarly, Searle concludes, a computer executing the program would not understand the conversation either.

So - what might this program consist of?  Obviously there is no simple algorithm for taking in a string of Chinese characters one by one, and sending out a meaningful response character by character.  It would need all sorts of features, such as memory of the current conversation (to provide context to any given input), ability to distinguish questions from comments from opinions, and so much more.  Of course any such program could never be carried out in a step by step manual way by a person, unless you are willing to wait days if not months or years for responses!

If we simply assume that such a program exists and works as described, then it does seem to me that the outsider interacting with the room would grant a level of understanding to it.  The Watson program that can play Jeopardy seems to be getting relatively close to this level of sophistication, although it was built for the Answer/Question format only.