Implicit learning ability is distinct from IQ or working memory

Tuesday, July 16th, 2019

Our implicit learning ability seems to be distinct from IQ or working memory:

Priya Kalra at the University of Wisconsin-Madison and her colleagues gave 64 healthy young adults four types of tasks that required implicit learning. One involved detecting an artificial grammar (after studying a series of letter strings that all adhered to undisclosed grammatical rules, the participants had to judge which strings among a new set were “grammatical” and which were not.) The second task required them to learn whether a particular group of images was going to trigger one outcome, or another (and they were given feedback to help them to learn). For the third task, they had to predict where a circle was going to appear on a screen, based on prior experience, during which the circle sometimes appeared in a predictable sequence of positions and sometimes did not. Finally, they had to learn visual categories implicitly: with the help of feedback, they had to classify abstract visual stimuli into one of two categories. (Explicit learning could have fed into some of these tasks, but the researchers made efforts to investigate, and take into account, its contribution for each individual.)

One week later, the participants returned to complete different versions of all these tasks, as well as tests of working memory, explicit learning (they had to deliberately learn a list of words) and IQ.

For three of the four implicit learning tasks, the researchers found a “medium” level relationship between a participant’s initial performance and how well they did a week later. This suggests stability in implicit learning ability. The exception was the artificial grammar task; the researchers think it’s possible that explicit learning “contaminated” implicit learning in this task at the second time-point.

The team also found that how good a participant was at implicit learning bore no relation to their IQ or working memory results. It seems, then, to be driven by independent neural processes to those that underpin explicit learning, which is linked to IQ.

This finding fits with earlier work that has tied explicit and implicit learning to different brain regions and networks. (The hippocampus is important for explicit but not implicit earning, for example, whereas damage to the basal ganglia and cerebellum impair implicit, but not explicit, learning.)

The new results have implications for theories that intelligence depends upon a single fundamental factor, such as processing speed, the researchers write. “These data … provide evidence for the existence of a completely uncorrelated cognitive ability,” they added.

The findings also imply that someone might feasibly be smart, as measured by an IQ test, but poorer at implicit learning than someone else with a significantly lower IQ score.

This might explain all kinds of things, from book-smart nerds with no common sense, to Jared Diamond’s acquaintances in Papua New Guinea.


  1. Ross says:

    Sounds similar to parts of the analytic section of the GRE.

  2. Wang Wei Lin says:

    Given feedback? Is that the same as leading the witness? What kind of ‘survey’ gives feedback? Seems you could ‘feedback’ you way to the desired outcome. Trash study.

  3. Felix says:

    For those interested, it appears that some of the input for the last of these “implicit learning” tasks is at:

    The paper seems to be behind the Elsevier paywall.

  4. Aretae says:

    I’m fascinated but unconvinced at least on the uncorrelated side.

    As you know, I focus on implicit leaning models … research like this is super interesting.

  5. Kirk says:

    This is pretty much in perfect alignment with what I’ve been saying for years: IQ testing really and fundamentally only tests the ability to do well on an IQ test, and your attempted application of that information outside the realm of the IQ test is going to produce some really wildly variable results…

Leave a Reply