So I’m thinking about starting something science-y. A friend of mine and I wanted to start a science blog last year, partly as a way of staying in touch and partly because we thought it would be fun, but it never came together. Now I’m back on that idea, only slightly modified. I’ve been thinking about how the fundamental principals of science are based on replication. You don’t get any theories until you’ve repeated the experiment many times. In psychology, though, you don’t see that happening very often. Everyone is so interested in coming up with the new idea, so you see lots of people making chimps listen to rap music in an fMRI machine, or having people play video games while on a unicycle in an fMRI machine, or giving mice tiny little lightsabres to fight each other inside a miniature fMRI machine. You don’t, however, see a second person verifying the newly famous Stanford Giving Mice Tiny Little Lightsabres to Fight Each Other Inside a Miniature fMRI Machine Experiment. I think it’s very important to do so, and it’s falling by the wayside.
Failure to replicate or attempt to replicate means that some experiments can produce an effect because of a quirk of the testing environment, and then that effect can become an assumption for someone else’s experiment and never be questioned. Things like this can happen.
Another big problem with this is that people use fMRI experiments to localize high-level brain functions, and then other folks use those results to figure out what else is going on in even higher levels of brain function. Lets say you play a bunch of tones at people and you figure out where the representation for the diminished fifth is. You get a bunch of people to do this, do a bunch of stats, and report that it’s in the parietal lobe between the area that controls scratching yourself and the area that thinks about Jessica Alba*. You publish, and 200 people read your paper. Of those 200 people, 20 get past the abstract, and 4 of those people aren’t just reading it to look at pretty brain pictures. One of those people decides to do a study about music and uses your paper as a reference. So now we’ve got someone looking in the parietal lobe for music interpretation. They isolate the parietal lobe for experimentation to save time, and they run some stats using a program someone else designed. Of the 1000 ways they could have interpreted the massively complex data they’ve collected, they decide to go with the one that says that when you play Black Sabbath to a macaque in an fMRI machine, their parietal lobe gets really excited when you also give them a ham sandwich to eat. And lo, Parietal Activation of Macaques Eating Ham Sandwiches and Tripping Balls To Black Sabbath is published in a reputable journal, and now we have two papers confirming the location of music in the parietal lobe. This continues indefinitely, and somewhere in Latvia, a little orphan cries himself to sleep, not knowing why.
Anyhoo, my idea is that I want to run a blog where I pick one recent non-fMRI-related study each week to turn into a web experiment using the program I wrote over the past two years. It’s quite sophisticated and surprisingly easy to set up, so it would be able to handle some pretty complicated experiments with ease. I would collect responses from about 100 people each time (if it manages to become wildly famous, that is), and then post the results a week later. Either that, or they’d last for a few weeks and overlap. In general, though, I’d want to start and end a study every week. I’d aim for about 20% completely new studies that I’d be curious about, and 80% recycled studies to replicate. People would visit the site if they were bored, and they’d be able to sign in under anonymous usernames and never have to type in their age and gender and all that other crap more than once. Data could be connected from study to study, and mined ruthlessly for all sorts of interesting effects. People would waste hours and hours on this site, all in the name of Science. Stats would be calculated on the fly for me to quickly post, and every study would be catalogued and rated based on whether their results were confirmed. Best of all, I would do it in the style of a Mythbusters kind of thing, and I’d get to slam BUSTED on any theory that didn’t hold up to Internet muster. I think this would be very cool. Because Science–as we all know via Bill Nye–RULES.
I’m currently welcoming names for this blog.
* The Parieto-occipital Jessica Alba Sulcus is, however, widely confirmed by many neuroanatomical studies.
2 Comments
My brain doesn’t feel psychological right now and I think I’m burned out from all the work I did yesterday, so I’m doing some house hunting. I’m looking on craigslist, and all the apartments I’ve come across are pretty much the same with the exception of price. I’m most interested in location, so I’m making a google map of all the places that interest me.
I think a really fleshed out house hunting google mashup would be so useful in this situation. I’m not just talking about taking a feed from craigslist and a few other apartment rental websites and overlaying them on a google map. I’m talking about having information on every possible thing that could interest a house hunter. Here are some ideas:
- Noise. This would be a combination of noise complaint information gathered by the Toronto Police, and external sources that track noise from traffic, industry, hustle bustle, etc. It would be neat if cell phone providers could collect information on the amount of background noise in phone calls along with which cell tower they’re connected to, but obviously that would be a violation of privacy unless people opted in, which definitely wouldn’t happen. It would also be neat if companies responsible for measuring work safety all sent noise level data to a central area to construct a map of average noise across the GTA. The Beau sent me this data from Transport Canada projecting noise from airplanes. (BTW, if you need any info about airplanes, he’s your guy in the sky.) There are many ways you could get this sort of data, and I know I would find it invaluable for choosing a place to live.
- Crime. The Toronto Star is supposed to have a mashup of homicide stats and grow-ops, but the homicide one doesn’t seem to be working at the moment. This Spot Crime site is useful, too. I would have though Scarborough would be more dangerous than it appears on this map. Knowing about homicides may be useful, but they can be pretty rare and pretty random. They also have a very high vividness, so even hearing about one murder (like the execution-style murder that happened at a grow-op down the street from us) could make you avoid an otherwise completely safe area. I’d be interested in muggings, rapes and bicycle thefts (although not necessarily in that order). The best way to display this data would be some sort of per capita probability, so denser areas don’t automatically look like more dangerous areas.
- Schools and employment. I’m not in this boat yet, but knowing the level of education and possibilities for employment in a particular area would be useful for new families. You could easily include EQAO stats for schools in your areas of interest (that shows how well schools are doing on standardized tests), average income for a particular area or what industry sectors are most active.
- TTC Routes and Timing. This one is obvious. Knowing about access to public transportation (especially for someone without a car, like me) would make a big difference to a lot of people. Knowing how long a particular trip would take would be even more useful. And since the TTC have been slacking off on their Trip Planner for years, I say it’s up to us to figure this out. All we need is for everyone in the city to log in and tell a site how long their morning and afternoon trips take. You can bet there will also be a couple TTC obsessives who would put in their data every single day. A few algorithms later, and kapow! We have realistic estimates for bus routes across the city, and also across time.
More ideas as I think of them.
1 Comment
In response to JTL’s recent findings on male-female relationships, I would like to introduce an invention that I thought up on Tuesday when I was at a country bar. It is called the non-engagement engagement ring, and you wear it on the ring finger of your left hand when you’re going out to a club and you don’t want to get hit on. It’s kinda like one of those celtic heart ring thingies, only repurposed without the new age religious overtones. Also, it’s way less lame.
One of the benefits of the non-engagement engagement ring is that you can personalize the ring to fit whichever kind of club you’re going to. So if you’re going to a classy bar it would be a diamond ring or maybe one of those lockets where you keep your cocaine, and if you’re going to a punk club it would be filled with anthrax. (Do not confuse the two when you leave your house to go to the bar.) If you’re going to a rave, it would be day-glo pink and in the shape of a soother, or in my case at a country bar you could wear a stylish bit of rope or a ring with “my baby’s mama” or “born cuntry” engraved into it. Of course, if you’re going to a gay bar there’s a whole other system to adhere to. The possibilities are endless.
If you are a boy you may not think this invention is necessary, but a girl gets VERY TIRED of having to use the “oh yeah, I do that with myboyfriend” line every time a guy flirts with her, especially in situations where it feels very ditzy to do so. (Boy: “I agree that it’s a slippery slope from the weak AI hypothesis, but I wouldn’t call the The Chinese Room Experiment bullshit.” Girl: “My boyfriend has big muscles.”) A non-engagement engagement ring would solve this completely. All the girl would have to do to get her point across (subtly but unambiguously and without embarrassment to the boy), is to take a drink with her left hand. If the guy is too nervous to make his interests known, the problem is taken care of, and if he’s actually sincerely interested in the conversation, there’s none of the awkwardness involved in a pre-emptive strike. It’s win-win!
1 Comment
Any runner worth his or her salt needs to check out Gmaps Pedometer. It allows you to plan out a running route using Google Maps, giving you information about the distance travelled, your elevation at each point, and optionally how many calories you’d burn.
I want to get back into running, so I’m excited that I no longer have to go through exhausting calculations with maps and rulers in order to figure out a new route. The last time I planned it using a ruler, I measured my route as 10km when it was actually 8km, which severely threw off my training schedule. Now I can be exact!
It would be a cool feature to allow users to type in a distance they want to travel and then click a button to automatically generate a random route within certain given constraints. You could have an option to optimize the route for “interestingness” or “prettiness”; users indicate places on their map where they want to walk past or where it’s really pretty and it’s planned accordingly. You could aggregate responses from all users to develop a topographic “prettiness” map of the city, allowing the system to suggest other pretty places based on user recommendations.
1 Comment
How easy would it be to combine PDP++ with some sort of audio statistical/Fourier analysis to teach it to respond to emotion in music / speech?
Side quote:
“An endeavor to imitate the involuntary modulations of the voice, and make its recitation richer and more expressive, may therefore possibly have led our ancestors to the discovery of the first means of musical expression.”
– Helmholtz
Comment
Darnit, darnit, darnit!
I don’t think Spielberg needs to be worried, but this film-making robot seems to have all the makings of a decent director.
It uses a neural network to develop its own appreciation of well-composed and aesthetically pleasing shots. And, unlike some human filmmakers I could mention, it displays an aversion to using scenes similar to anything used before.
They’ve made my idea. Yes, my idea.
Darnit!
Comment
I just wrote in an email to Bob McDonald at Quirks & Quarks. I really hope he answers my question on the radio. Here it is:
Hi Bob,
I’ve got a question about something that happened to me (particularly my brain) a few weeks ago. I tend (for some inexplicable reason) to faint in the shower if it gets too warm, and one time I had a very strange experience. I felt myself beginning to lose consciousness, so I left the shower and sat down for a while to get a bit of fresh air and recover. Usually, I see spots that move around a bit, which I assume is the result of lowered activity or overactivity due to blood loss in the part of my brain that controls vision. But then, something completely unexpected happened: I started to see in 8-bit colour! Any idea what could have caused this?
Cheers, Eve
Explanations welcome.
5 Comments