Sunday, December 20, 2009


Do some words grab your attention more than others because of their semantic content? If I want to get the attention of 12 screaming kids, would I be better off yelling "SEX!" or "EGGPLANT!"

This was the topic (kinda) of a study recently reviewed by the excellent Cognitive Daily blog: Huang, Y., Baddeley, A., & Young, A. (2008). Attentional capture by emotional stimuli is modulated by semantic processing. Journal of Experimental Psychology: Human Perception and Performance, 34 (2), 328-339 DOI: 10.1037/0096-1523.34.2.328.

The study used an interesting methodology: rapid serial visual presentation, or RSVP which involves showing participants a random stream of stimuli, flashing by one every tenth of a second. Wiz bang! That's a lot of flashing. Let Cognitive Daily explain:

Typically if you're asked to spot two items in an RSVP presentation, you'll miss the second one if it occurs between about 2/10 and 4/10 of a second after the first one, but not sooner or later. This phenomenon is called Attentional Blink -- a blind spot caused by the temporary distraction of seeing the first item... Their streams were simply random strings of letters and digits, with two words embedded in each stream. Then they asked students to look for words naming fruit as they flashed by. If a fruit word appeared, it was always the second word in a stream. The key was in the first word: half the time, this first word was a neutral word like bus, vest, bowl, tool, elbow, or tower, and half the time it was an emotional word like rape, grief, torture, failure or morgue. So a sequence might look like this:

  • JW34KA
  • QPLX12
  • MC15KW
  • 083FLB
  • S21L0C
  • DJW09S
  • 3LW8Z9
  • XOWL01
And so on. The first word acts as a distractor: the students are looking for fruit words, but this is always a non-fruit word. The question is, are emotional words more distracting?

The results result of the experiments was ...

 a qualified yes. When the participants were asked to pay attention to the meaning of the words (e.g., "look for words that mean fruit"), then yes, there was a distractor effect (i.e., participants were less accurate at identifying the fruit words at the relevant lag; they simply flashed by without being recognized). However, when asked to perform a different task, like "look for words that are all caps," then no, there was no effect. From the author's abstract:

Only when semantic processing of stimuli was required did emotional distractors capture more attention than neutral distractors and increase attentional blink magnitude. Combining the results from 5 experiments, the authors co=]nclude that semantic processing can modulate the attentional capture effect of emotional stimuli.

The original paper is behind PsycNET's firewall so I don't have access to it, but of course, my curiosity is piqued. I recall that the time course of visual word recognition was no simple thing. This task requires participants to recognize words and perform a decision about the words at a very rapid pace. Trying to tease apart what's occurring in the word recognition process during this would probably fill a dissertation, or at least a really good series of publishable papers.

Also, how did they determine what counted as an emotional word? Was some kind of experiment performed whereby participants were shown a series of words and their blood pressure was measured? EEG? fMRI? Galvanic skin response? What then? How does one determine that one out-of-context lexical item has more emotional effect than another? There may be good research on this, I don't know. But it seems intuitive that context has a lot to do with our emotional response to meaning. We know that torture was one of the words. But what is its emotional effect in the following context: I love my kids, but watching Barney with them is torture. They also used failure. Really? That has enough of a predictable emotional effect to be used in an experiment like this? Why only negative words? Why not wealthy, powerful, gorgeous? Did they use the word moist? They should've used moist.

Even if some words can be shown to have significant and predictable emotional effects all by themselves, these effects could easily be mitigated by the experimental design. For this experiment, they showed 16 participants 128 sequences. That's a lot of rapid flashing. If the word torture is in the 120th sequence, I don't think I'm necessarily going to be processing the full range of semantic associations anymore. I'm going to be doing the minimal amount of linguistic work necessary. As a participant, I will, in essence, be gaming the system.

But then again, psycholinguistics is tough. I respect anyone who explores new paradigms for studying something that is, currently, impossible to see: how language works in the brain. But by the same token, it's healthy to put these methods through rigorous debate (if a blog post can be counted as rigorous debate...).

BTW, there are some nifty online demos for Attentional Blink methodology. Enjoy:
PS: on a humor note. My original example to contrast with "SEX!" was "TOMATO!", then I changed it to "EGGPLANT!" Eggplant just seems funnier. I resisted the urge to go with the obvious kumquat (when deciding which spelling to use, I Googled kumquat and discovered that it truly is the devil's fruit; it had 666,000 hits).

1 comment:

public sex positions said...
This comment has been removed by a blog administrator.

Putting the Linguistics into Kaggle Competitions

In the spirit of Dr. Emily Bender’s NAACL blog post Putting the Linguistics in Computational Linguistics , I want to apply some of her thou...