Cognitive Daily

Take a look at these two images. Do they belong in the same category or different categories?


You say the same? Wrong — they’re different! The one on the right is a little blurrier.

What about these two?


These are in the same category. Sure, the one on the right is still blurrier, but now it’s rotated a bit, so that puts the two objects back in the same category. My rule for categorizing is complex, involving both blurriness and rotation (I’ll explain how it works later on).

How do you think you would do if you were tested on these categories? Do you think you’d do better or worse if you were put in a high pressure situation? Suppose you were being tested along with a partner, and if each of you could categorize 80 percent of the items correctly, then you’d both get a $10 bonus. Suppose your partner had already passed the test, so now your score would determine if you and your partner would get the bonus. Would you choke under pressure?

It turns out, with a test like this, you’d probably do better. But in an easier test, where you were only asked to categorize based on a single rule such as blurriness or rotation, you’d do worse under pressure. That’s what a team led by Arthur Markman discovered in a recent study. Inspired in part by research we’ve discussed on CogDaily showing that people with larger working memory are more likely to choke under pressure, the team hypothesized that people would also be less likely to choke under pressure on tasks that don’t rely so much on working memory.

Paradoxically, the more complex categorization tasks rely less on working memory. Take a look at this graph:


The green dot represents the figure on the left in the two examples above. In graph a, categories are defined solely based on sharpness, so blurrier images are in one category, and sharper images are in the other. In graph b, categories depend on both sharpness and rotation, so an equally blurry image can be in two different categories, depending on rotation. That’s how the second image, both blurrier and rotated compared to the first image, manages to be in the same category. This type of categorization is too difficult to process with working memory, Markman’s team argues, so it’s processed using “information-integration.”

Pressure to perform well places demands on working memory, but not on information-integration, so we perform better on this sort of task when we’re under pressure. In the experiment, participants were placed into two groups — high-pressure and low-pressure. The high-pressure group was told the story about their partner passing the test and both partners’ rewards depending on their own performance. The low-pressure group was simply asked to do the best they could. Half of each group was given an information-integration task where two rules had to be applied simultaneously. The other half was given the simpler, single rule-based task. Here are the results:


The high-pressure group performed significantly better than the low-pressure group on the information-integration task, but choked compared to the low pressure group on the rule-based task. Markman et al. argue that this result demonstrates that we can excel under pressure at tasks which don’t rely on working memory.

These results suggest something about how to approach high-pressure situations in general: try to limit the demands on your working memory when you’re in high-pressure situations. Do you have any practical ideas on how to limit use of working memory? Let us know in the comments.

Markman, A.B., Maddox, W.T., & Worthy, D.A. (2006). Choking and excelling under pressure. Psychological Science, 17(11), 944-948.


  1. #1 Harlan
    December 21, 2006

    Just a note that Dave’s showing response accuracy in the final block of training. That 2% difference in the rule-based condition was not statistically significant, but there was a statistically significant reduction in learning in the high pressure group overall as they learned the rule-based problem. So the punch line is the same…

    I think this work is also consistent with research in expertise in stressful situations. People are good in stressful tasks when they can rely on well-trained skills that don’t require much working memory. If they have to think on the fly, though, stress can be a big problem, reducing creativity.

    Interesting work, as always, from Markman’s lab…

  2. #2 Dave Munger
    December 21, 2006

    Yeah, I was definitely simplifying things here just a bit. Fascinating stuff, though.

  3. #3 Jegger
    December 21, 2006

    This seems to be a validation for one of the basic tenets of the “Getting Things Done” system: “Get it out of your head”. By writing down everything that needs to be done, you limit what’s in your working memory. Also, by making sure that you break things down into actual physical actions, you keep yourself from having to create an ad-hoc plan and store it in working memory.

  4. #4 schizo
    December 21, 2006

    Performance anxiety brings to my mind my patients who need counseling because they cannot, well, perform. Wonder if it is because the working memory required during sex is high. Is it?

  5. #5 Chris
    December 22, 2006

    It might help to note that Maddox has been doing work for several years in which he’s shown that there are two distinct categorization systems for these types of perceptual categories, and that research was the basis for this experiment. The reason the rule-based categorization system suffers under pressure is because it’s largely conscious and universal, and therefore subject to cognitive load, working memory constraints, attentional limitations, etc., whereas the non-rule based system categorizes automatically, and unconsciously, and is therefore largely effortless.

New comments have been disabled.