Sunday, February 27, 2011

Weekend Update

I know, it's been over a week since I've updated my blog! Well, I have been very busy most of the week, and the rest of the time, I admit, I've been hanging out with Jana.

Well, let's see... Last Sunday I was able to write 1000 words for my literature review, and then Monday and Tuesday I ran subjects in lab all day. Tuesday was my longest day, as I tacked on 3 and a half hours of data analysis with my adviser to discover from 20 raw scores that no hemispheric difference existed! It sucked, but we decided that Gabors may not be attentionally-demanding enough to produce differences-- either that, or we hadn't used the right design.

So we spent half the evening coming up with different experiments for the second part, and went home with nothing substantial. Adviser asked me to turn the results into d'prime just in case something turned up significant. d'prime shows a ratio for each person's scores for how many times they said they saw a Gabor when it was actually there, compared to how many times they said they saw a Gabor when it wasn't there. This is hits versus false alarms, just another measure for significant differences.

Well, I spent a large part of Wednesday doing that, and a good thing too-- because it looks like turning the scores into d'prime shows an effect! Again, not an effect of hemisphere, but an effect of set size-- let me explain.

In the original task I asked subjects to perform, they had to look at the center of a computer screen where a rapid stream of letters appeared. They had to pick out a white letter X among the stream of black letters, and then, sometime after the white letter appeared, a circle of 12 gabors appeared around the central letter stream. I think I've told you before that Gabors are funny little stripey-oval-things that are made up of different frequencies of the background color-- a luminance pattern, if you will-- here, I'll show a picture...



OK, that's not my picture-- but that's what the display looks like-- except for the letters in the corner, that just demonstrates that a white letter appears in the center among black letters. So Gabors are those things in the circle around the letter stream. And as you can see, there is an odd-one-out at an angle to the others. Not only did people have to quickly determine if the white letter that appears is an X, but they also have to pick out the odd-one-out.

If you split the circle of Gabors into slots of 3 (top 3, bottom 3, right 3, and left 3, total of 12), we only had the odd-one-out Gabor appear randomly in either the left or right 3 slots, to examine any effect of side (side differences equate hemispheric differences).

Well, odd-one-out detection does not require sophisticated search-- it just pops out at you. This is automatic feature detection. However, if the letter stream and circle of Gabors go by fast enough (80 milliseconds for each letter, and 80 milliseconds for the Gabor display), then the pop-out Gabor is harder to catch. This is an especially hard task if the Gabors appear at the exact same time as the white letter-- this is because you can't pay attention to 2 things at once, so when you're focusing on finding the white letter, you don't have time to disengage your attention to the periphery to see the pop-out Gabor. However, if the Gabors appear 800 milliseconds after the white letter, that gives your attention plenty of time to disengage from the white letter, and prepare yourself for pop-out detection.

So in my experiment, we only found this effect of time between white letter/Gabor onset. No effect for whether the Gabor appeared on the right side or the left side of the display, which is what we were hoping to find.

HOWEVER... After I ran subjects in the 12-Gabor condition, we thought it would be interesting to see if people showed attentional differences if there were only 6 Gabors on the screen ("set size" of 6). So for the last half of subjects, we took out the top and bottom 3 Gabors (which were just distractors, after all) and kept in the 3 slots on either side in which the odd-one-out could appear. Now this is interesting-- why should people show any difference in performance when the placement of the odd-one-out Gabor didn't change? All we did was remove some Gabors that subjects never even had to attend to.

Well! Using d'prime, we found that in the 6-Gabor task, people show a significantly smaller difference between correct hits and false alarms. Which means they were worse at correctly identifying a pop-out that was there, and/or they made more errors by indicating they saw a Gabor when it wasn't actually there, than in the 12-Gabor task.

How can this be? you may be asking! Since people are looking at fewer objects, shouldn't the 6-Gabor task be easier? Listen up-- here's how it goes-- when people have 12 Gabors, and 11 of them are all at the same angle, they look very unified, like they're all supposed to be at that angle. So when an odd-one-out shows up, it interrupts the perfect circle. But if there are only 6 Gabors, there are only 5 at the same angle, and an one odd-one-out makes the display look like a mess and less like an actual odd one among a uniform group. Thus, it is harder to spot. And that is what we found.

Yes! We found a big different in d'prime between the 12-Gabor and 6-Gabor set size. Hurray Hurray for us.

BUT- I STILL want to see some hemispheric differences! Which is why I'm excited to move on to phase 2-- the conjunction task.

You see, single features like Gabors can be seen as pop-outs-- they don't require sophisticated search. But if you combine features, and ask subjects to looks for a combination of features (called conjunctions), then the task depends on much greater attention!

Therefore, if we change the task so that the odd-one-out is a letter instead of a Gabor-- and letters are composed of different combinations of features (lines/etc.)-- then we should not only find an opposite effect of set size (conjunctions should be easier to spot in a set size of 6 because it is less to search through), but we should find hemispheric differences as well (because the right hemisphere is better at attention, and in this task, attention needs to really work hard).

So that's what I've been doing all week. Not only have we been planning the next phase of the experiment, but I'm also scrambling for participants to run a control next week (where they only have to look for the odd-one-out Gabor and ignore the letter stream-- which should be easy since they're not switching attention from one thing to another).

On top of that, I have to get "random experiment" for my course up and running. Yes, I'm still working on that. If you remember, I decided to study the effects of fantasy proneness on confusion of words and pictures. See, if people have to look at a series of words and pictures, I hypothesized that fantasy-prone individuals are more likely to forget what was a word and what was a picture. I was going to study hallucination-prone individuals, because all I wanted to do was replicate an experiment done using hallucinating and non-hallucinating schizophrenics, but with normal people. But I didn't have time to wait for the ethics to clear and apparently asking people if they've ever hallucinated is not the most ethical thing to do. So I went with fantasy-proneness instead, seeing as fantasy-proneness is highly correlated with hallucination-proneness, so my high-fantasy-prone people SHOULD also be hallucinators... I just can't ask them that.

Anyway, we'll see what happens. I've already analyzed the initial survey and I have made my groups of high- and low-fantasy-prone people. I just need to contact those people and create timeslots for them to come into the lab to run the actual experiment. First, I have to make sure the experiment I created in e-prime (a REALLY EASY computer program that puts timed visual stimuli together for you) works on HOC's lab computers. I will do that tomorrow. Then, after class, I will post the timeslots and hopefully get that experiment out of the way by the end of this week or next.

As for my poster for BPS, I have completely restructured it, and I'm happier with it now than I was. I was ABOUT to submit it-- I had one version in PDF format and everything-- but I couldn't bring myself to do it because I was ashamed of how it looked. So I had Jana go over it and she helped me change the color scheme/etc, and it looks SO much nicer now. I just have to change a FEW things here and there, and it'll be ready to print. I should be able to print it sometime next week.

Anyway, I'll end this post here-- I still have to talk about social things I've done this week, like how I went out to dinner with friends on my course, and I'm sure you want to hear about the hiking club, or what Jana and I have been up to outside of class, or anything that DOESN'T have to do with coursework. But unfortunately, you guessed it-- I'm doing COURSEWORK today-- I have 500 words left to write for my weekend quota, and I've got to get on it cos Jana wants to watch movies tonight.

So I will be sure to update on normal bits of life AFTER my 500 words is finished, and if I have time before Jana texts me: 'U come over, good time.'

No comments:

Post a Comment