When Control Hurts: Better Rule Learning in Subjects With Lower Memory Span

The ability to actively maintain more information in memory, known as "working memory," seems to benefit performance in a variety of tasks. One idea is that these tasks require controlled attention, allowing for better control over behavior.

But there's a serious problem with this explanation: maybe this doesn't reflect improved control so much as superior motivation. In other words, maybe subjects with higher working memory are the only ones who care, and everyone else is just goofing off!

Thankfully, there are some cases where additional working memory has no benefit - or can even be a disadvantage. A great example of this can be found in Decaro et al's recent article in Cognition, who show that subjects with higher working memory capacity are actually worse at learning complex rules in a categorization task.

To demonstrate this, Decaro et al. asked 71 subjects to complete four categorization tasks: half were based on categorizing stimuli according to simple rules involving one dimension (e.g., if the dimension were shape, category A might contain all circular items, and category B all squarish items) and half were based on more complex rules (e.g., category A should contain items which satisfy any two or more of three possible constraints, for example squareness, redness, or largeness).

Subjects were given 200 trials to learn each rule, and were provided with feedback on each trial. They also rated how much pressure they felt to learn the task - to ensure that any individual differences were not simply motivational in nature. Subjects also completed 2 measures of working memory: operation span (in which subjects have to remember certain letters while they are completing various arithmetic problems), and reading span (in which subjects have to remember certain letters while they read various sentences).

The surprising result: subjects who were able to remember more letters on these tasks actually took longer to learn the complex categorization rule! (Here, "learning" is defined as the number of trials until they got 8 trials correct consecutively). This result was reversed for the simpler rule, such that subjects with better working memory learned that rule more quickly. Furthermore, memory scores were unrelated to perceived pressure, providing preliminary evidence that higher scores are not merely related to motivation.

Why should this occur? Some theorists have proposed that sustained control or effort can actually impair the acquisition of habits. Decaro et al. propose that the complex categorization task requires exactly this form of habitual learning - which is clearly impaired in subjects who appear to have more controlled resources at their disposal.

There are a few other cases where those with "better" cognitive capacity actually perform worse on specific cognitive tasks. One is the famous matchstick arithmetic problem, where subjects with brain damage frequently outperform healthy subjects. Similarly, some children who suffer from febrile convulsions can actually show superior performance relative to healthy children. Likewise, kids can outperform adults, and chimps may outperform humans at some tests of short-term memory. And rats genetically altered so that they can't produce new neurons will actually show improved memory in some tasks.

An interesting question: do all the "brain training" programs which are beginning to show true cognitive effects actually make it harder for subjects to learn these types of categorization tasks? Or does trained attention/memory not succumb to the same pitfalls as more "natural" dispositions towards greater memory and attention?