darb.ketyov.com

Caveat lector: This blog is where I try out new ideas. I will often be wrong, but that's the point.

Home | Personal | Entertainment | Professional | Publications | Blog

Search Archive

15.12.13

Brain Log? Collaborative academic neuroscience blog

Earlier today on Twitter I mentioned that I feel that academic neuroscience could use a collaborative, professional blog such as Language Log. This seemed to resonate with several people, but 140 characters isn't enough to explain why I feel this way.

First and foremost: for those of you not familiar with Language Log, it's a blog run by academic Linguists and other language professionals, including several journalists who cover language.

Off the top of my head, here are a few reasons why I think neuroscience could use a "Brain Log":

  • Language Log is collaborative, meaning it doesn't "belong" to any one person.
  • This minimizes one issue of a lot of personal blogs: personal "branding" and link-bait style posts. The focus is on solving problems rather than general interestingness.
  • Many neuroscience blogs are fantastic, and have higher level discussions (e.g., Neuroskeptic), but they're run on a professional site and thus are for-pay content. And they're still "branded" by their individual ownership, leaving further discussions at the discretion of the owner.
  • The scope of Language Log is more technical, which allows the people "in the trenches" to dig into details that the general public may not find very interesting, but young researchers, journalists, etc. may find very useful as a resource.
  • A collaborative blog allows for better two-way discussion between the primary scientists and journalists covering neuroscience (a very hot topic in the public interest right now).
  • This would allow journalists better access to broader scientific opinions, and to give them an easy way to reach out to academics.

11.12.13

An intuitive explanation of over-fitting

A question over on Quora piqued my interest: What is an intuitive explanation of over-fitting? I use this blog and public writing to try and explain neuroscience topics, but I don't think I've ever taken a whack at explaining statistics. Which now strikes me as strange considering how much of my research is computational and methods-focused. So I gave it a go. Let me know if this makes sense.

Imagine you've got two normally distributed random variables, x and y. Here is their relationship:

When you know all of the data, it should be clear that knowledge of the value of x provides little information about y. That is, there is no clear predictive relationship. Certainly not a linear one.

We can test this by trying to fit a linear model (y = ax + b) to the data (the red line). This procedure shows us that the variance in x accounts for less than 1% of the variance in y.

But let's imagine you can't collect all of the data. Instead your sample includes only two of those data points. Again you try and fit a linear model and...
Holy crap! Your linear model (y = ax + b) explains 100% of the variance! Now you might say, "my linear model is perfect and completely explains the relationship between x and y". But wait... that fitting estimate doesn't look the same as the linear fit when we've got the full data:
In fact, our "perfect" model is pretty terrible!

That is overfitting. When your model has too many parameters relative to the number of data points, you're prone to overestimate the utility of your model.

We can keep going with this:
When your sample contains three data points, our linear model (y = ax + b) once again performs poorly (~7% of the variance) but our new model, a quadratic model (y = ax^2 + bx + c) explains 100% of the variance.

Again, sampling one more data point:
Well now our linear and quadratic models both perform poorly, but our new new model, a cubic model (y = ax^3 + bx^2 + cx + d), explains 100% of the variance.

And so on.

6.12.13

DIY Brain Stimulation

A few months ago I wrote a piece for the BBC on the foc.us do-it-yourself, consumer-based brain stimulator. Here's the article in full; full text follows:

For centuries people have sought quick fixes and miracle cures to enhance our vitality and make us faster, stronger, smarter, younger. Historically speaking, many of these "fixes" have been pharmacological. Freud used cocaine to improve his mood and treat his patients while Queen Victoria was known to enjoy a wine infused with the drug. In fact it's only relatively recently – the late 1960s – that the sale of amphetamines was made illegal.

Now it seems the next form of mental stimulant could come in the shape of a piece of technology, and not a drug. Looking like something Star Trek's Borg might wear, the foc.us device promises to "make your synapses fire faster" with the application of a small electrical current to the brain for 30 minutes. The promised result is that you'll have sharpened reflexes and faster reactions. As a gamer, I can see the attraction. Those extra split seconds can make all the difference. It's certainly a shrewd business decision too: video games were a £43 billion ($67 billion) dollar industry in 2012, and professional gaming draws in sponsorship and viewing figures that rival physical sports. The company's bold claims may draw in others looking for a mental boost, and not just people who like to play video games.



Unfortunately as a scientist I'm sceptical. The technique being used by the foc.us headset is known as transcranial direct current stimulation (tDCS). In the scientific world, studies using it have actually been quite promising. In carefully controlled experiments, low-level electrical excitation of just the right brain regions have provided a small boost to task performance. After a quick zap, volunteers have gotten slightly better at noticing if a square changed colour (after an hour of watching flashing coloured squares). But could this really extend to a task as complex as playing a video game?

The general idea is that if a certain brain region is involved in cognition – attention for example – then stimulating that area with an electrical pulse (priming it) should improve your ability to attend to what's happening around you. We also know that when you're engaged in a complex cognitive task (like playing Call Of Duty: Modern Warfare), certain neurones in specific brain areas fire more quickly.

So, I imagine the thinking at foc.us is that if we can speed up the rate of neuronal firing we should be able to enhance gaming ability. Unfortunately, that's a little like saying, "I can boost my computer's memory by jolting its hard drive with a battery!" – your IT department won't be happy.

Yes, sometimes tDCS provides a cognitive benefit; the data is indisputable. I myself make use of this technology in my lab. But its capabilities are too non- specific. In the brain, it matters which neurones are excited; alcohol excites a lot of neurones, but no one would claim that drinking four pints sharpens your reflexes.

And while stimulation methods have been proven to be safe for use in controlled experimental situations over the short- term, we don't know about long–term effects. We've been down this road before. Over the short term the data was very clear: cocaine and amphetamines were very effective at enhancing energy and mood, but we didn't know about long-term negative consequences. I will never say that we shouldn't test new drugs or new technology, but when companies promise to "excite your prefrontal cortex" to give you "the edge in online gaming" and "let the force of electricity excite your neurones into firing faster", well... buyer beware.